Psychophysical Measurement of Neural Adaptation Abnormalities in Magnocellular and Parvocellular Pathways in Glaucoma

Abstract

purpose. It is well established that contrast sensitivity is reduced in glaucoma. This study explored whether such contrast processing abnormalities consist of an absolute threshold level difference or a problem with contrast gain control.

methods. Seventeen patients with primary open-angle glaucoma and 17 approximately age-matched control subjects participated. Subjects were tested foveally and midperipherally (12.5°). Subjects with glaucoma were tested in a peripheral region of relatively normal visual field (neighboring locations required to be within the normal 95% confidence limit on the total deviation plot of their most recent SITA/full threshold Humphrey Field Analyzer assessment; Carl Zeiss Meditec, Dublin, CA). Control subjects were tested in matching locations. Contrast discrimination was assessed using the steady-pedestal (magnocellular [M] pathway) and pulsed-pedestal (parvocellular [P] pathway) stimuli of Pokorny and Smith for seven pedestal luminances between 15 and 75 cd/m², presented on a background of 30 cd/m².

results. Glaucoma group thresholds were significantly elevated compared with control subjects foveally and peripherally on both the pulsed-pedestal (P) and steady-pedestal (M) tasks (P < 0.01). Effect size statistics revealed slightly greater deficits on the P pathway task and greater deficits for pedestals that were decrements, rather than increments, from the surround luminance. Foveal deficits were of a magnitude to be explained by a reduction in contrast sensitivity; however, the peripheral deficits were greater than predicted by this factor alone.

conclusions. Foveal and midperipheral dysfunction of both M and P pathways was identified in people with glaucoma, in areas of relatively normal visual field performance. These findings are supportive of nonselective neural adaptation abnormalities in early glaucoma.

Perimetric tasks that evaluate the performance of specific ganglion cell types have been shown to be superior to standard automated perimetry (SAP) for the detection of early glaucomatous visual field loss (for example, see Refs. ^{1 2 3 4 5 6} ). Examples of visual-function–specific tasks include those that preferentially assess the magnocellular (M) pathways (such as frequency doubling technology Perimetry, FDT) or the koniocellular (K) pathways (such as short-wavelength automated perimetry, SWAP). These clinical tests, which isolate function of a particular subset of retinal ganglion cells, have been widely investigated, with the purpose of finding better methods for detecting and monitoring glaucomatous visual field loss. 

Although these newer perimetric test strategies have been clinically effective, ^{1 2 3 4 5 6} there has been conjecture regarding the underlying reasons for their relative advantage over SAP for the detection of early loss. Two main theories have been proposed. The first of these theories proposes that neurons with larger cell bodies and axon diameters are at greater risk early in the disease process, and hence are selectively damaged. ^{7 8} As both M and K neurons are larger than parvocellular (P) neurons, ^{9 10} this hypothesis predicts selective loss of M and K function early in disease. Early histologic reports support the loss of larger ganglion cells in glaucoma ^{7 8} ; however, more recent reports do not. ^{11 12 13} Furthermore, whereas M and K cells are larger on average than P cells, the P cell body and axon diameter increases in the peripheral retina, resulting in substantial overlap in morphology. ^{9 10}  

The second theory proposed to explain the earlier loss found with perimetric tasks that measure either M or K pathway function proposes that loss is more readily detected in these pathways because they have minimal neural redundancy. ¹⁴ There are far fewer M and K neurons than P (the retinal ganglion cell population comprises approximately 80% P, 10% M, and 5%–10% K), resulting in a sparse retinal distribution. ^{9 10 15} This theory proposes that even if neurons were lost proportionally across all cell types, the sparser systems would demonstrate functional loss earlier due to this reduced redundancy. More recent histologic reports provide support for nonselective neural loss in glaucoma. ¹³  

Although studies exploring the clinical utility of selective functional assessment of M and K pathways are numerous (for example, see Refs. ^{1 2 3 4 5 6} ), there are fewer studies that have been designed to assess P pathway function selectively. Given that histologic evidence for degeneration of the P layer of the lateral geniculate nucleus in experimental glaucoma is compelling (for review see Ref. ¹⁶ ), psychophysically measuring P cell function should prove useful in assessing glaucomatous damage. There are some reports of deficits on tasks presumably mediated by the P pathways. These include red–green chromatic tasks, ^{17 18} high-resolution tasks such as high-pass resolution perimetry (HRP), ¹⁹ and those measuring spatial aliasing. ²⁰ In the main, it has been difficult to compare the magnitude of deficits between M and P tasks directly, because the procedures used to measure function in these subpopulations often differ markedly, not only in stimulus composition but also in retinal locations tested and thresholding methodologies used (for example: FDT vs. HRP). 

Extensive exploration of visual-function–specific tests has been motivated by a need to find better measures of early functional loss due to glaucoma. A nonlinear relationship between the number of ganglion cells and SAP performance has been demonstrated in primate models of glaucoma, so that a 6-dB (mild) loss in visual field sensitivity may correspond to up to 30% to 50% ganglion cell loss. ²¹ It would be ideal to be able to measure not only deficits arising from cell death early in the disease process, but also those arising from cell malfunction (sick cells). Measurement of abnormal cell function before death may be of particular importance for assessing the potential efficacy of neuroprotective strategies. One possible avenue for measuring abnormal function may be to use tasks that assess neural adaptation. It is possible that “sick” cells are in altered adaptational states, resulting in adaptation occurring more slowly, or incompletely, or so that light and dark adaptation are affected differentially. There is some evidence for adaptational abnormalities in early glaucoma, ²² and alterations in metabolism in both M and P neurons have been demonstrated in primates with experimentally induced glaucoma. ²³ If so, visual-function–specific tasks in which performance is dependent at least in part on retinal adaptation mechanisms, may be advantageous in detecting early glaucomatous loss. 

The purpose of this study was to compare functional performance in early glaucoma using a task that selectively assesses M and P pathway function and that requires normal neural adaptation. We used the contrast-discrimination tasks (steady-pedestal and pulsed-pedestal) described by Pokorny and Smith ²⁴ that we have used previously to identify contrast processing abnormalities in people with migraine. ²⁵ These tasks briefly present luminance increments on luminance pedestals. With the steady-pedestal stimulus, the subject must make a contrast-discrimination judgment after adapting to the pedestal luminance, whereas with the pulsed pedestal, the subject adapts to the background. These tasks are ideal to address our purposes as the test stimulus used to measure M and P function is identical, only the prestimulus adaptation phase differs. We were interested in exploring two questions: whether greater loss can be identified using the contrast-discrimination paradigm than is predicted by contrast sensitivity and whether there is evidence for different degrees of loss in the M and P pathways. 

Methods

Subjects

Seventeen subjects with primary open-angle glaucoma (POAG) and 17 control subjects participated. The subjects were approximately but not exactly age-matched, and there was no significant difference (t-test: t ₍₃₂₎ = −1.45, P = 0.16) in mean age between the groups (mean ± SD age in the glaucoma group, 69.7 ± 9.6 years; mean ± SD age in the control group, 65.6 ± 7.5 years). Subjects with glaucoma were recruited from the Glaucoma Clinic of the Lions Eye Institute (Perth, Australia) and were all under the care of a glaucoma specialist ophthalmologist (one of the authors, WHM). Control subjects were recruited by written advertisement in a local newspaper or were friends or family of the participants with glaucoma. 

Subjects with glaucoma were required to have a clinical diagnosis of POAG and a previously documented glaucomatous visual field loss (glaucoma hemifield test [GHT] results outside normal limits or pattern standard deviation indices worse than the 5% probability level) as established with a Humphrey Field Analyzer (HFA) 24-2 or 30-2 full threshold or SITA procedure (Carl Zeiss Meditec, Dublin, CA) in the eye to be tested. Visual field deficits ranged from early to more advanced visual field loss (mean defect [MD] ranged from 0.17 to –17.45 dB, mean, −7.45 dB; pattern standard deviation ranged from 1.91 to 15.11 dB, mean, 9.59 dB). 

Both control and glaucoma subjects were required to be free from other eye disease, to be free from systemic disease known to affect visual function, to have best corrected visual acuity of 20/25 or better, and to have refractive errors no greater than ±5.00 D sphere with no more than 2.00 D cylinder. Control subjects were also required to have intraocular pressure of less than 21 mm Hg measured with applanation tonometry. For control subjects, eligibility to participate was determined by a comprehensive eye examination as part of the study. Eligibility for patients with glaucoma was determined from the results of their most recent regular visit to the ophthalmologist. 

Before participation, all subjects provided written informed consent in accordance with a protocol approved by the University of Western Australia Human Research Ethics Committee, and in accordance with the tenets of the Declaration of Helsinki. 

Steady-Pedestal and Pulsed-Pedestal Contrast-Discrimination Stimuli

The test stimuli were based on those described by Pokorny and Smith. ²⁴ Stimuli were generated using a video card (VSG 2/5; Cambridge Research Systems, Kent, UK) housed in a 933-MHz computer. The software was a custom-designed program (MatLab, ver. 6.1; The MathWorks, Natick, MA) that we have used previously. ²⁵ Stimuli were displayed on a γ-corrected, high-intensity, gray-scale monitor (frame rate 100 Hz, CIE1931 x: 0.25, y: 0.3; model GD 402; Phillips, Eindhoven, The Netherlands) which was viewed at a distance of 75 cm (the monitor subtended 28.5° × 21.5°). A chin and forehead rest was used. After each trial, the subject’s responses were collected with a button box (model CB3; Cambridge Research Systems). 

Figure 1 shows the contrast-discrimination stimuli. The steady-pedestal condition is shown in Figure 1A . For this condition, four squares (the pedestal) are presented continuously, within a 30-cd/m² surround. Subjects adapt to the steady pedestal for 1 minute before commencing the run, and the pedestal was presented continuously during the interstimulus interval (3 seconds). During the brief (30 ms) test interval, one of the squares was incremented in luminance, and the observer was instructed to indicate the location of the increment. Subjects chose whether the briefly incremented brighter square was on the left or right side and therefore made a two-alternate, forced-choice (2AFC) decision. For foveal testing, each of the four squares were 1° of visual angle, separated by 9 min arc. The black fixation dot was 9 min arc and was presented in the center of the screen. 

Figure 1B shows the pulsed-pedestal condition. During the adapting phase, a small fixation marker was presented continuously within the 30-cd/m² surround. The four squares were presented only during the brief test interval (30 ms), with three of the squares having the pedestal luminance and one square having the pedestal in addition to a luminance increment. The subject’s task was the same as for the steady-pedestal task. The stimulus presented during the test interval was the same for both tasks, as was the timing. It was only the presence or absence of the pedestal squares during the adapting phase that differed. The same pedestal luminances were used for both pulsed and steady conditions and included both decrements (pedestal squares of 15 and 24 cd/m²) and increments (pedestal squares of 38, 60, and 75 cd/m²) from the mean luminance (30 cd/m²). Performance was also measured for a pedestal luminance equal to that of the background (30 cd/m²), which effectively provided a measure of contrast sensitivity for the single test square alone. 

Peripheral Stimuli

Subjects were tested both foveally and at a single peripheral test location (the choice of the location is described subsequently). Figure 1C shows the placement of stimuli for peripheral testing. Stimuli were presented on the diagonal meridians, with squares placed so that the nearest corner to fixation was at 10° and the center of the stimulus was at 12.5° from fixation. A 2AFC procedure was used, in which the two possible test squares were those on the diagonals indicated by the crosses in Figure 1C . Diagonally opposed squares were chosen so that the two test squares could be presented at the same retinal eccentricity. 

The size of the squares was increased to 1.73°, and the gap between the squares was increased to 13 min arc for peripheral testing. The area of the squares was increased to enable the assessment of approximately equal numbers of ganglion cells foveally and peripherally. Estimates of ganglion cell density as a function of eccentricity were obtained from Figure 6 of Curcio and Allen. ¹⁵ Because the estimated ratio of M and P cells varies only slightly from fovea to 12.5°, we did not additionally scale the stimuli for this factor. Estimates of M and P cells ratios, determined using equations 1 and 2 from Wang et al., ²⁶ predicted the percentage of P cells to decrease from approximately 95% at the fovea to 93% at 12.5° and the ratio of M cells to increase from approximately 5% to 7%. 

Choice of Peripheral Test Location

Participants with glaucoma were tested in a quadrant of normal or near normal performance on their most recent HFA test. The test quadrant required visual field performance that was not flagged as abnormal on the total deviation plot in locations adjacent to the test area, which was centered on 12.5°. An example is shown in Figure 2 . 

HFA visual fields were not measured specifically for this study, but as part of the subject’s regular glaucoma management. As some of the HFA fields were a few months old, visual fields were measured on the day of testing, by using a perimeter with the central threshold test (model M700; Medmont Pty Ltd., Camberwell, Australia). A detailed description of the perimeter can be found elsewhere. ²⁷ In brief, the perimeter uses 0.43° (Goldman size III) light-emitting diodes (LEDs) as stimuli (λ_max = 565 nm). The bowl luminance is 3.2 cd/m² and the maximum stimulus luminance is 320 cd/m². The central threshold test uses a ZEST thresholding algorithm and thresholds 99 visual field locations arranged in concentric rings (3°, 6°, 10°, 15°, 22°, and 30°). The Medmont perimeter returns visual fields comparable to the HFA, ²⁸ and these fields were used to confirm the test quadrant. The Medmont central threshold test also returns two global indices: the average defect (AD) and pattern defect (PD). These indices provide a summary of generalized performance across the visual field (AD) and localized asymmetry (PD) and are similar to the total deviation and pattern standard deviation indices of the HFA. Control subjects also underwent visual field assessment, and all had normal SAP fields (Medmont central threshold test, global indices; P > 0.05). Control subjects were tested in the same eye and peripheral location as their approximately age-matched counterpart with glaucoma. 

Thresholding Procedure

Thresholds were determined using a three-down, one-up staircase (79% correct performance level) ²⁹ whereby subjects were required to make three correct responses for the luminance increment to be decreased by 20%, or one incorrect response for it to be increased by 20%. Staircases commenced at a luminance increment level that was clearly visible and terminated after six reversals. The threshold was determined as the mean of the last four reversals. No feedback regarding the accuracy of responses was provided. Foveal and peripheral measures were obtained in separate runs. 

Testing was divided into two test sessions of approximately 45 minutes. The longest duration between first and second test visit was 1 month. Foveal testing was performed at the first visit and peripheral testing at the second. Subjects were randomly assigned so that half performed the steady pedestal condition first. 

Statistics

Glaucoma and control group performance was compared by using parametric statistics (t-test, or ANOVA) after confirming that the distributions of results were not significantly different from Gaussian (Kolmogorov-Smirnov test, P > 0.05). Statistical analysis was performed on computer (SigmaStat 3.0; SPSS Science, Chicago, IL). 

Effect sizes ³⁰ were determined to enable comparison of the magnitude of deficits across the different tasks. Effect sizes represent the difference between the groups in the number of standard deviations. Effect size (d) was calculated as  

\[d\ {=}\ ({\mu}_{\mathrm{g}}\ {-}\ {\mu}_{\mathrm{c}})/{\varsigma}_{\mathrm{pooled}}\]

where  

\[{\varsigma}_{\mathrm{pooled}}\ {=}\ {\surd}{[}{\varsigma}_{\mathrm{g}}^{2}\ {+}\ {\varsigma}_{\mathrm{c}}^{2})/2{]}\]

and μ_g and μ_c are the glaucoma and control group means, and ς_g and ς_c are the standard deviations, respectively. 

Results

Contrast-Discrimination Thresholds: Raw Data

Figure 3 shows the performances of both subject groups for the steady- and pulsed-pedestal conditions. Group mean (± SE) performance is shown. The contrast-discrimination functions for the steady- and pulsed-pedestal stimuli differed markedly, as expected. ²⁴ The steady-pedestal condition resulted in a linear relationship between log threshold and log pedestal luminance, whereas the pulsed-pedestal condition resulted in a characteristic V-shaped curve. ²⁴ The function obtained for the steady condition was monotonic for pedestals that were either increments or decrements from the surround, indicating local adaptation to the pedestal luminance. For the pulsed condition, thresholds were increased as pedestal luminances were either increased or decreased in intensity from the surround. In this case, the contrast difference between the pedestal and the background influenced performance, rather than adaptation to the pedestal, as in the steady condition. 

To verify that the shape of the contrast-discrimination functions were consistent with those previously reported for this method, curves were fit to the data using equations previously shown to adequately describe such contrast-discrimination data. ²⁴ The steady-pedestal data were fit using linear regression which resulted in slopes of approximately 1.0 as expected ²⁴ (95% confidence intervals for slope: control foveal = 0.9–1.6; control peripheral = 0.9–1.5; glaucoma foveal = 0.8–1.2; glaucoma peripheral = 0.8–1.4). The pulsed-pedestal data were fit using equation 3 from Pokorny and Smith ²⁴ :  

\[{\Delta}C\ {=}\frac{K(10/R_{\mathrm{max}})(C_{\mathrm{sat}}\ {+}\ C)^{2}}{C_{\mathrm{sat}}\ {-}\ (10/R_{\mathrm{max}})(C_{\mathrm{sat}}\ {+}\ C)}\]

where ΔC is the contrast-discrimination threshold, R _max is the maximum response amplitude, C _sat is the semisaturation constant (the contrast at which the response amplitude is half R _max), C is the Weber contrast, and K is a vertical scaling parameter. As in Pokorny and Smith, C _sat was set equal to 1.0; R _max and K were free parameters in the curve fit; and luminance difference rather than ΔC was plotted in Figure 3 . The percentage of contrast gain is determined as (R _max/C _sat)/100. The values of R _max determined from the best curve fits to the data in Figure 3 ranged from 20 to 40, which are similar to those measured previously ^{24 25} for this test stimulus, and result in estimates of percentage contrast gain within the typical range for the P pathway. ³¹  

Inspection of Figure 3 reveals that the glaucoma group’s mean performance was worse than that of the control group for both steady and pulsed conditions. Two-way, repeated-measures ANOVA (factors were group and pedestal) demonstrated that the subjects with glaucoma performed significantly worse than control subjects at both retinal eccentricities, for both steady-pedestal (fovea: F_(32,1) = 15.0; P < 0.001; periphery: F_{(32, 1)} = 22.94, P < 0.001) and pulsed-pedestal (fovea: F_(32,1) = 19.35; P < 0.001; periphery: F_(32,1) = 49.54; P < 0.001) tasks. For the steady-pedestal condition, there was no significant interaction between group and pedestal luminance for either foveal or eccentric viewing. For the pulsed-pedestal task, Figure 3 shows greater differential performance between glaucoma and control groups for pedestals that were decrements from the background than increments. This interaction between group and pedestal luminance was statistically significant (P = 0.01) both foveally and peripherally. 

The results presented thus far show that the glaucoma group performed worse than the control group on both pedestal tasks, indicative of both M and P pathway dysfunction. Effect sizes were determined to establish whether there was a relatively greater loss of either M or P pathway performance. Effect sizes were calculated for each pedestal luminance, and then an average measure was determined separately for decremental and incremental pedestals (Table 1) . When the pedestal was equivalent to the background, the steady- and pulsed-pedestal conditions converged. This condition became a contrast sensitivity measure for the single test square, and the effect size for this is listed in the table as “no pedestal.” Table 1 shows similar magnitudes of loss for the putative M and P tasks once variability on the tasks is taken into consideration. 

Comparison of Contrast-Discrimination Performance to Perimetric Performance

To visualize performance as a function of disease severity, Figure 4 shows scatterplots of individual subjects’ contrast thresholds for a pedestal luminance of 24 cd/m², plotted against the global indices returned by the Medmont perimeter on the day of testing. Medmont global indices are shown, rather than those from the HFA, as some of the HFA fields conducted as part of the glaucoma subjects’ routine ophthalmic care were several months old at the time of contrast-discrimination testing. Figure 4 shows performance for the 24-cd/m² pedestal, as this was the pedestal luminance that showed greatest separation in performance between control subjects and participants with glaucoma (Fig. 3) . Figure 4 shows that performance on the contrast-discrimination tasks was largely unrelated to perimetric indices. This may be expected, as all subjects, including those with more advanced glaucoma, were tested in an area of visual field that was spared. 

Figure 4 also enables inspection of the relative difference between performance on the steady- and pulsed-pedestal tasks. It is possible that the difference between the M and P pathway performance may vary as a function of disease severity (in this case measured by visual field deficit severity). Inspection of Figure 4 reveals no obvious trend for differential involvement of M and P pathway performance as a function of visual field severity. To further explore this question we calculated for each subject the average difference between putative P (pulsed-pedestal) and M (steady-pedestal) performance. The difference between thresholds measured with pulsed and steady pedestals was determined for each pedestal luminance and the average calculated. This is referred to as the P–M difference. We then calculated Pearson product moment correlations between the P–M differences and the Medmont visual field global indices (Table 2) . None of these correlations was statistically significant, indicating no significant trend for the relative difference between P and M pathways to be more or less exacerbated for a particular disease severity. 

We also calculated average P–M differences for each of the control subjects and compared the distribution of these differences with those of the glaucoma subjects. The control and glaucoma groups were not significantly different for either foveal (t ₍₃₂₎ = 0.30, P = 0.77) or peripheral (t ₍₃₂₎ = 1.38, P = 0.18) measures. This demonstrates that the relative difference between P and M pathway performance was similar in the glaucoma and control groups and is supportive of a nonselective loss of both M and P function in the glaucoma group. 

Comparison of Group Performance when Normalized for Contrast Sensitivity

It has been established that both contrast sensitivity ^{32 33} and spatial localization ^{34 35} are impaired in people with glaucoma. Although we tested in areas of relatively normal visual fields, it is possible that the increased thresholds in the glaucoma group represent reduced contrast sensitivity and that no additional advantage is conferred by the pedestal task. The pedestal tasks also required subjects to localize the stimulus spatially. Although this is unlikely to create difficulties in the fovea, because the test stimulus separation is large, in the periphery it is possible that an impaired ability to localize the stimulus accurately would contribute to poor performance. The condition in which the pedestal luminance is equivalent to the background luminance is effectively a measure of contrast sensitivity for the single test square and includes the same localization judgment as when the pedestal is present. To investigate the contribution of the pedestal to the decreased performance of the glaucomatous observers, we normalized each subject’s data by dividing the thresholds at each pedestal luminance by the threshold for the single square alone (the condition in which the pedestal luminance was equivalent to the background luminance). 

Figure 5 shows normalized performance for foveal viewing, and it can be seen that there was no significant difference between the control and glaucoma groups (two-way, repeated-measures ANOVA: pulsed-pedestal F_(32,1) = 0.41, P = 0.53; steady-pedestal F_(32,1) = 0.77, P = 0.39). Hence, the difference in foveal performance found before normalization can be explained by differences in the ability to perceive the test square in the absence of an additional pedestal. 

Contrast sensitivity normalized performance for the midperipheral stimulus is shown in Figure 6 . The glaucoma group still performed more poorly than the control group at all pedestal luminances, and the difference is statistically significant for both pulsed (F_(31,1) = 7.49, P = 0.01) and steady (F_(31,1) = 4.2, P = 0.05) tasks. The greatest magnitude of loss was present for the pulsed-pedestal task when the pedestals were decrements from the background luminance (effect sizes: pulsed-pedestal decrements = 1.16, increments = 0.46; steady-pedestal decrements = 0.72, increments = 0.47). Hence, for the peripheral task, a greater loss was measured than can be explained by either a reduction in contrast sensitivity to the test square alone or by the ability to localize the test square in the midperipheral visual field. 

Discussion

This study found evidence for functional loss of information transmitted by both M and P pathways in POAG. The magnitude of the deficit was similar in the two pathways, and we found no evidence for a selective loss of function in either pathway. This is consistent with recent histologic evidence. ¹³  

The test stimuli used in this study have been well studied ^{24 36 37} and have been shown to have adaptational, contrast gain and temporal summation properties consistent with those of the M and P pathways in normal observers. We were able to fit our data adequately with the same equations that fit normal data, which suggests that in our elderly and glaucomatous patients, the task is still likely to measure the separate performance of the M and P pathways. 

After contrast sensitivity normalization, we found deficits consistent with a genuine contrast adaptational deficit in the midperipheral visual field that could not be explained either by the subject’s ability to see the test square or to identify its location. Foveal performance was no different from that of control subjects after normalization. Both foveal and peripheral areas were chosen to have normal visual field performance assessed with standard automated perimetry. Although glaucomatous visual field loss can take many forms, it is typical for glaucomatous loss to be measurable first outside the fovea. ³⁸ All our subjects had midperipheral loss in at least one other quadrant and thus demonstrated this typical pattern. Hence, it may be expected that greater functional loss would be measurable peripherally than foveally. Indeed, had we found loss of similar magnitude both foveally and peripherally, the deficits might have been explained by lack of attention in the presence of the pedestal or other processes unrelated to ganglion cell function. That both M and P cells become larger and more sparsely distributed in the peripheral visual field ¹⁵ may have contributed to our enhanced ability to assess functional loss in the midperiphery. 

Glaucomatous deficits were more pronounced for pedestal luminances that were decrements rather than increments from the background luminance. M and P ganglion cells are further subdivided into ON- and OFF-center types that are thought to enable the perception of light increments and decrements. ³⁹ The dendritic fields of ON-center parasol (M) and midget (P) cells are 30% to 50% larger in diameter than the corresponding OFF-center cells. ¹⁰ Differential psychophysical performance in relation to decrements and increments has been reported in a variety of tasks, ^{40 41 42} providing support for ON-OFF functional asymmetry. Further exploration is needed to confirm whether OFF-pathway assessment yields benefits in measuring early glaucomatous loss. 

The findings of this study imply that neural adaptational abnormalities are present in early glaucoma. In the midperipheral field, the residual glaucomatous deficit remaining after our normalization procedure is consistent with a genuine contrast adaptational deficit that could not be explained either by the subjects’ ability to see the test square or identify its location. There is some support in the literature for abnormal foveal adaptation in glaucoma, ²² and it is conceivable that neural malfunctioning ²³ would result in poor adaptational control before cell death. Further study into neural adaptation and contrast gain abnormalities in glaucoma seems warranted and may have the potential to provide enhanced measures of early glaucomatous damage. 

 Supported in part by Raine Medical Research Foundation Priming Grant (AMM), a National Health Medical Research Council (NHMRC) Program grant (WHM), and the McCusker Glaucoma Centre. AMM is supported by NHMRC Australian Clinical Research Fellowship Grant 139150.

 Submitted for publication November 11, 2003; revised January 6, 2004; accepted January 15, 2004.

 Disclosure: A.M. McKendrick, None; D.R. Badcock, None; W.H. Morgan, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Illustration of the contrast-discrimination stimuli. (A) In the steady-pedestal condition, a black fixation dot was presented at the center of the continuously displayed array of four squares. During the test interval (30 ms) one of the squares was incremented in luminance. For the pulsed-pedestal condition (B), the black fixation dot was presented within the adapting field (30 cd/m²), and during the test interval the four-square array was presented briefly (30 ms), with one of the squares incremented in luminance relative to the other three; (C) Schematic of stimulus positioning for peripheral testing. The stimulus comprising four squares was presented on the diagonal meridians within in a single quadrant (chosen to be a quadrant with normal performance on the TD probability plot for patients with glaucoma and matched for the control subjects). Subjects fixated a marker in the center of the screen. Stimuli were placed so that the center of the four squares was 12.5° from fixation.

Figure 2.

 

Sample right eye visual field (HFA, SITA Standard 24-2) from one glaucomatous participant (aged 66) illustrating placement of the peripheral stimuli in an area of relatively normal performance (within the normative 95% confidence limits for the total deviation plot).

Figure 3.

 

Performance on the contrast-discrimination tasks in both subject groups. Data are presented as the mean ± SE for the control (open symbols) and glaucoma (filled symbols) groups. Curves through the circles represent the best-fitting linear regression line. Curves through the triangles are the least-squares best fit of equation 3 from Pokorny and Smith. ²⁴ Vertical dotted line: condition in which the pedestal luminance equaled the background luminance (i.e., the pedestal was invisible). Pedestals that were luminance decrements from the background are shown in the shaded area and increments in the nonshaded area.

Table 1.

 

Effect Sizes that Describe the Magnitude of the Difference between the Control and Glaucoma Groups Relative to the Spread of Performance within the Groups

	Steady Pedestal (Putative M)					Pulsed Pedestal (Putative P)
		Decrements	No Pedestal	Increments	Decrements	Increments
Fovea	1.14	1.27	0.87	1.26	0.68
Periphery	1.60	0.68	1.36	1.88	1.53

Figure 4.

 

Comparison of visual field performance to contrast-discrimination performance. Shown are scatterplots of individual glaucoma subjects’ contrast-discrimination thresholds (for the 24 cd/m² pedestal condition) plotted against the Medmont central threshold test global indices. (A, C) Average defect; (B, D) pattern defect.

Table 2.

 

Pearson Product Moment Correlation Coefficients between the Individual Glaucoma Subject Visual Field Global Indices and Their P–M Contrast Threshold Difference

	Fovea	Periphery
Average Defect	R = 0.42, P = 0.10	R =−0.22, P = 0.39
Pattern Defect	R = 0.06, P = 0.81	R =−0.34, P = 0.17

Figure 5.

 

Normalized foveal performance on the contrast-discrimination tasks. Data are presented as the mean ± SE for the control (open symbols) and glaucoma (filled symbols) groups. Vertical dotted line: condition in which the pedestal luminance equaled the background luminance (i.e., the pedestal was invisible). Individual performance for this condition was used to normalize the data.

Figure 6.

 

Normalized midperipheral performance on the contrast-discrimination tasks. Data are presented as the mean ± SE for the control (open symbols) and glaucoma (filled symbols) groups. Vertical dotted line: condition in which the pedestal luminance equaled the background luminance (i.e., the pedestal was invisible). Individual performance for this condition was used to normalize the data.