Psychophysical Evidence for Rod Vulnerability in Age-Related Macular Degeneration

Abstract

purpose. To determine whether there is rod system dysfunction in the central retina of patients with age-related macular degeneration (AMD).

methods. Dark-adapted sensitivity (500-nm stimulus) and light-adapted sensitivity (600 nm) were measured psychophysically at 52 loci in the central 38° (diameter) of retina in 80 patients with AMD, and results were compared with those from older adult normal controls. All dark-adapted data were corrected for preretinal absorption.

results. Mean field dark-adapted sensitivity was significantly lower in AMD patients as a group than in normal subjects. Within the AMD group were subsets of patients with normal mean dark- and light-adapted sensitivities; reduced dark-adapted sensitivities without detectable light-adapted losses; both types of losses; and, least commonly, only light-adapted losses. Regional retinal analyses of the dark-adapted deficit indicated the greatest severity was 2° to 4° or approximately 1 mm from the fovea, and the deficit decreased with increasing eccentricity.

conclusions. These psychophysical results are consistent with histopathologic findings of a selective vulnerability for parafoveal rod photoreceptors in AMD. The different patterns of rod and cone system losses among patients at similar clinical stages reinforces the notion that AMD is a group of disorders with underlying heterogeneity of mechanism of visual loss. Dark-adapted macula-wide testing may be a useful complement to the more traditional outcome measures of fundus pathology and foveal cone-based psychophysics in future AMD trials.

Age-related macular degeneration (AMD) is a heterogeneous group of disorders in which older adults lose central retinal photoreceptors, either by an atrophic process, the most common disease expression, or by a neovascular event, the more destructive form causing severe central vision loss. ¹ The mechanisms underlying vision loss in AMD are likely to be multifactorial, undoubtedly complex, and are incompletely understood. ^{2 3} Recent histopathologic and morphometric studies of human donor retinas with AMD indicate a predilection for parafoveal loss of rods over cones in the early, nonexudative form of the disease. ⁴ Although both rods and cones in the parafovea degenerate in early AMD, rod loss precedes and is more severe than cone loss in most of the donor retinas evaluated. Even in the exudative form of AMD, there is greater retention of cones compared to rods. ⁴  

Prompted by these histopathologic observations and earlier studies indicating abnormal vision in AMD under dark-adapted or low luminance conditions, ^{5 6 7 8 9 10 11 12 13 14 15} we examined the hypothesis that there is vulnerability of rods early in these conditions using psychophysical tests of dark-adapted visual function. If indeed this is the case, these tests may serve as useful assays for evaluating disease progression and the effectiveness of treatments targeted at early AMD pathogenesis. 

Methods

Subjects

Patients with AMD were recruited from the Retina and Vitreous Service of the Department of Ophthalmology, University of Alabama (Birmingham), and the Scheie Eye Institute, University of Pennsylvania (Philadelphia). Inclusion criteria were as follows: (1) at least 55 years old; (2) 20/60 visual acuity or better (best-corrected) in the eye to be tested psychophysically; and (3) a diagnosis of AMD in the test eye. 

Fundus photographs were evaluated with a macular grading scale based on the international classification and grading system ¹⁶ and other scales that have been described. ^{17 18} AMD was subclassified into early and late forms. ¹⁹ Seventy-one patients qualified as early AMD, defined as having at least five large drusen (>63 μm) with or without focal hyperpigmentation. Nine patients had late AMD, defined as having choroidal neovascularization or geographic atrophy (>175 μm in diameter): three patients had choroidal neovascularization (2 extrafoveal and 1 subfoveal, <1 disc diameter) and six had geographic atrophy (single or multiple atrophic foci mainly in the para- and perifoveal region). Fundus pathology in the eye tested with psychophysics was further characterized in the early AMD patients by estimating the percent of retina covered by large drusen within the central 3000-μm diameter area. For this purpose, 66 of 71 photographs were used; 5 of lesser quality were excluded. Drusen coverage was categorized as follows: <10%; 10% to <25%; 25% to<50%; 50% to <75%; and >75% of retinal area. Fundus photographs from the fellow eyes of most of the patients were available and evaluated; patients were subcategorized as having either bilateral large drusen or large drusen in the test eye and a unilateral disciform scar in the fellow eye. 

Exclusion criteria for the AMD sample were as follows: (1) glaucoma, ocular hypertension, diabetes, or any other ocular, neurologic, or systemic disease that would compromise vision in either eye, as indicated by a comprehensive eye examination within 6 months of enrollment; and (2) use of medications that would complicate interpretation of the data (e.g., retinotoxic drugs). The final sample of AMD patients consisted of 80 individuals with mean age 74.5 ± 6.6 years old (mean ± SD; range, 59–91 years). There were 44 women and 36 men (99% white, 1% African American). 

Older adult control subjects (n = 12; 4 women and 8 men, all of whom were white) also were in this study. Mean age of this normal control group was 71.3 ± 5.0 years old (range, 62–80 years). Fundus photographs of these subjects indicated either a normal fundus background appearance or the presence (<20) of hard drusen (<63μ m). Inclusion and exclusion criteria were as above except that acuity in each eye had to be 20/30 or better with no diagnosis of AMD. Seven subjects had 20/20, two had 20/25, and three had 20/30; those with 20/25 and 20/30 had small cataracts that likely contributed to their acuity level. 

All subjects had a routine ocular examination (including fundus photography) and static threshold perimetry. Institutional approval of studies was obtained at both participating institutions and the tenets of the Declaration of Helsinki were followed. Informed consent was given by all subjects after the nature and purpose of the study were explained. 

Procedures

Perimetry.

Dark- and light-adapted static threshold perimetry was performed with a modified automated perimeter (Humphrey Field Analyzer; Humphrey Instruments, San Leandro, CA); details of the instrumentation and methods have been described. ^{20 21 22} The pupil of the test eye was dilated with tropicamide 1% and phenylephrine hydrochloride 2.5%. Thresholds were measured with a 4-dB/2-dB (dB, decibel) staircase bracketing procedure using narrow band (∼15 nm full-width half-maximum) stimuli (1.7° diameter, 200-ms duration). Orange (600 nm) stimuli were used in the light-adapted (10 cd·m⁻²) state and blue-green (500 nm) stimuli dark-adapted (≥40 minutes) at 51 extrafoveal loci in the central 38° (diameter) of the visual field. There were 35 loci on a 6° grid (12° temporal field not tested) and 16 additional loci at 2°, 4°, 8°, and 10° eccentricity along the horizontal and vertical meridia (Fig. 1) . Mean and SD of the 51 loci were calculated. Sensitivity loss at each locus was calculated as the difference between the measured sensitivity and the mean normal sensitivity at that locus. A measurement was defined as normal if within ±2 SD from the mean normal value. To analyze for regional variation in function, sensitivity losses were combined according to their eccentricity (Fig. 1) : ring 1 (2°); ring 2 (4°), ring 3 (6°), ring 4 (8 to 8.5°), ring 5 (10°), ring 6 (12°), and ring 7 (13–19°). Sensitivity to light was expressed on a logarithmic scale (10 dB is equal to 1 log₁₀ unit), where higher numbers represent better sensitivity (lower threshold). All 500-nm data were corrected for preretinal absorption as described below. 

Rod or Cone Mediation.

Dark-adapted thresholds measured with the 500-nm stimulus in the current work are mediated by the rod system in normal subjects. ^{20 21} The difference between rod- and cone-mediated absolute thresholds to this stimulus is greater than 30 dB at the perifovea and more peripheral loci ^{20 21} ; at the parafovea (2–4° eccentric), the difference is somewhat smaller and varies between 25 and 30 dB. In patients, losses of up to 25 dB in dark-adapted sensitivity refer to dysfunction of their rod system. Greater than 25-dB losses in patients represent the minimum rod system dysfunction present (e.g., a 40-dB dark-adapted sensitivity loss suggests a rod system dysfunction of 40 dB or greater). The majority of the dark-adapted sensitivity losses obtained in the AMD patients of this study were <25 dB and thus represented dysfunction of their rod system. Dark-adapted sensitivities were also measured at the fovea with the 500-nm stimulus (by fixating in the center of a diamond formed by four dim red light-emitting diodes ^{20 21} ). Because of the relatively large size (1.7° diameter) of the stimulus, normal foveal thresholds are mediated by the rod system; the difference between rod- and cone-mediated thresholds, however, can be small, ²³ thus complicating interpretation of the origins of foveal dark-adapted sensitivity losses. The white background used for light-adapted thresholds desensitizes the rod system; losses of light-adapted sensitivity to the 600-nm stimulus represent the dysfunction of the long/middle wavelength cone system. 

Preretinal Absorption.

It is known that preretinal absorption may contribute to loss of sensitivity to shorter wavelength stimuli, especially in older subjects. ^{24 25} We used a psychophysical method to estimate and compensate for age-related nonretinal changes, which are mostly due to yellowing of the lens. The method takes advantage of the difference between the scotopic sensitivity spectrum of a subject and the sensitivity spectrum of the rod photoreceptor. ^{22 26 27 28 29 30} To abbreviate the method, only short and middle wavelengths are used. If no preretinal absorption is present, the measured rod-mediated sensitivity difference between the two wavelengths should be equal to the difference in sensitivity of a rod photoreceptor. Preretinal absorption, mostly occurring at shorter wavelengths, increases the sensitivity difference between the two wavelengths. 

In the current work, radiometrically matched short (410 or 420 nm) and middle (560 nm) wavelength stimuli (1.7° diameter, 200-ms duration) were presented to the dilated and dark-adapted eye at 15° nasal field to avoid macular pigment and enhance rod participation. Preretinal absorption at the short wavelength was estimated after compensating for the difference in human rod photoreceptor absorbance at the appropriate wavelength. ³¹  

To estimate the preretinal absorption at 500 nm, the wavelength at which dark-adapted perimetry was performed, we used a model that describes the total lens absorption spectrum as the sum of an observer-independent spectrum, and a scaled observer-specific spectrum changing with age. ³² Recent results obtained from lenses of donor eyes showed close correspondence between this two-component model and experimentally determined transmittance spectra. ³³ The scale factor of the observer-specific lens absorption spectrum is obtained by first subtracting the effect of the observer-independent spectrum from the estimated lens absorption at the short wavelength and then dividing the result by the difference of the two wavelengths of the observer-specific absorbance spectrum. For pseudophakic patients (13 of 80 patients, 2 of 12 normal subjects), the preretinal absorption correction was not performed because of the inapplicability of the model and negligible absorption by intraocular lenses at 500 nm. ³⁴  

In this psychophysical paradigm, dark-adapted sensitivities to short and middle wavelength stimuli are assumed to be mediated by the rod system. ³⁵ At the middle wavelength (560 nm) and the retinal locus (15° nasal field) used, this assumption is violated when the dark-adapted sensitivity loss exceeds 25 dB. ^{20 21} In 6 of 80 patients an extrapolated value of preretinal absorption was used because either the rod-mediation assumption was violated or thresholds were not reliable. The extrapolated value was obtained from linear regression analysis applied to absorption correction of 96 subjects (61 patients, 35 normal subjects) against age (c = 0.02 + 0.068·age; c in dB, age in years; r ² = 0.475). 

The distribution of the preretinal absorption correction was similar in the two groups [F(1,75) = 0.55, P = 0.46]; for AMD subjects, the correction was 4.9 ± 1.6 dB (mean ± SD) compared with the value for the normal subjects, which was 5.3 ± 1.8 dB. 

Statistical Analyses.

Analyses comparing mean sensitivities and sensitivity losses between groups were performed using analysis of variance (ANOVA) techniques as computed by SAS STAT software. One-way ANOVA was used for comparing mean levels across patient groups. Differences in the regional variation within the test field between patient groups were evaluated by using a repeated-measures ANOVA and testing the interaction between the patient group and eccentricity. The associations between the gradings of the degree of drusen coverage of the macular area and the sensitivities and visual acuity were assessed with the partial Spearman correlation coefficients. 

Results

Mean field sensitivity (average of 51 extrafoveal test loci, Fig. 1 ) was used as a measure of overall function of the central retinal region we studied. Dark-adapted mean field sensitivity was 6.7 dB lower for AMD patients (42.4 ± 8.1 dB; mean ± SD) than for controls (49.1 ± 3.0 dB); the two populations (see boxplots in Fig. 2A ) were significantly different [F(1,90) = 8.03, P = 0.006]. The distribution of dark-adapted mean field sensitivities in the AMD patients (Fig. 2A) indicates that 42 (52%) patients have normal sensitivity while 38 (48%) patients show abnormally reduced results. With respect to light-adapted sensitivity (Fig. 2B) , AMD patients on average exhibited a 2.2-dB deficit compared with controls [17.9 ± 2.5 versus 20.1 ± 1.2 dB; F(1,90) = 8.71, P = 0.004]. Abnormal light-adapted sensitivity was present in 32 (40%) of the patients. Among the 44 patients with abnormal mean field function, 26 showed both reduced dark- and light-adapted sensitivity, 12 had only dark-adapted dysfunction and 6 had only light-adapted sensitivity losses. In 39 of these 44 patients, the magnitude of mean field dark-adapted sensitivity loss exceeded the magnitude of light-adapted sensitivity loss (data not shown). 

Patients with selective dark-adapted dysfunction had mean field sensitivity losses ranging from 6.2 to 14.6 dB. Inspection of gray scale maps of these patients indicated that some had relatively homogeneous losses across the test field, whereas others showed more concentrated loss in one or more regions. A plot of SD versus mean dark-adapted sensitivity loss (Fig. 3A ) illustrates the different degrees of variation in function. Larger variations appear to be associated with greater degrees of sensitivity loss. Two representative maps from patients with different degrees of dark-adapted sensitivity loss but relatively little variation across the field are illustrated (Figs. 3B 3C ; mean losses of 7.0 and 10.7 dB, respectively). Two patients with similar degrees of loss but large variations (Figs. 3D 3E ; mean losses of 8.1 and 14.9 dB, respectively) are also shown. The variability in the latter patients tended to be due to a single region of more severe loss surrounded by other less affected areas. All six patients with only reduced light-adapted sensitivity had relatively mild dysfunction; mean field sensitivity losses ranged from 2.5 to 4.4 dB. In four of six, the dysfunction was distributed uniformly across the field. 

Analyses of the entire data set and subsets were undertaken to determine whether there were regional variations in sensitivity loss (Fig. 4) . Dark-adapted sensitivity impairment exhibited statistically significant regional variation (Fig. 4A) . Dark-adapted sensitivity loss was on average 2.65 dB worse in the regions of field covered by rings 1 to 4 (2–8.5° eccentricity) compared to outside this area[ F(1,79) = 28.52, P = < 0.0001]. With respect to the seven eccentricity rings, dark-adapted sensitivity loss in the AMD group decreased with increasing eccentricity within a 19° radius [F(6,553) = 3.55, P = 0.002]. The proportion of patients with abnormal dark-adapted sensitivity also decreased from ∼35% at rings 1 and 2 to ∼26% at ring 7. The subset of patients with no mean field abnormalities showed a predilection for paracentral dysfunction but this was not statistically significant (ring 1, F(1,46) = 2.31, P = 0.14). Mean parafoveal (2°, ring 1) sensitivity loss was 3.5 dB greater than that at the fovea[ F(1,78) = 28.52, P = < 0.001]. Light-adapted sensitivity impairment using the seven eccentricity rings (Fig. 4B) did not show statistically significant regional variation[ F(6,553) = 1.66, P = 0.13]. The subset of patients without mean field abnormalities showed no significant paracentral light-adapted dysfunction [ring 1, F(1,46) = 3.74, P = 0.06]. Parafoveal and foveal light-adapted losses were not different[ F(1,77) = 1.04, P = 0.67]. 

Were dark- or light-adapted sensitivity results related to degree of fundus pathology in the patients? Of the nine late AMD patients, six had both dark- and light-adapted sensitivity losses, and two had only loss under the dark-adapted condition. One patient without detectable abnormalities across the test field had subfoveal choroidal neovascularization. Among the early AMD patients, the amount of large drusen in the central 3000-μm diameter macular area showed a weak negative correlation to the LogMAR visual acuities (r′ =− 0.26; P = 0.04; Partial Spearman Correlation Coefficient, adjusted for age) but was not related to light-adapted (r′ = −0.12; P = 0.35) or dark-adapted (r′ = 0.05; P = 0.72) foveal sensitivities. Extrafoveal (rings 1 and 2) light- and dark-adapted sensitivities were then compared with the amount of large drusen. There was no statistically significant correlation between drusen grade and sensitivity losses measured dark-adapted (r′ = −0.19; P = 0.12) or light-adapted (r′ = −0.04; P = 0.74). 

We also asked if macula-wide sensitivities in the test eye of patients with bilateral large drusen differed from those of patients with fellow eyes having disciform scars. Of the patients studied, 27 could be classified as having bilateral large drusen and 34 as having unilateral disciform scars. Patients with unilateral disciform scars in the fellow eye had greater mean field dark-adapted sensitivity loss in the test eye than those with bilateral large drusen [8.48 versus 4.18 dB; F(1,59) = 4.77, P = 0.03] and also greater mean field light-adapted sensitivity losses [2.49 versus 1.20 dB; F(1,59) = 6.17; P = 0.02]. 

Discussion

Our psychophysical results from the central retina of patients at relatively early stages of AMD showed there can be prominent dark-adapted dysfunction attributable to the rod system. More patients showed rod dysfunction than cone dysfunction and, in most patients, the magnitude of the rod dysfunction was greater than that of cone dysfunction. These results are thus consistent with histopathologic studies that indicate a predilection for rod over cone photoreceptor loss in AMD. ⁴ Quantitative comparisons of rod and cone sensitivity losses should be the goal of future experiments to confirm and extend our findings with dark- and light-adapted perimetry. Light-adapted sensitivities, such as were measured in the current work, may underestimate the loss of cone photoreceptor sensitivity, depending on the hypothesized mechanism of disease action. For example, if AMD simply causes a reduction of photopigment density in cone photoreceptors, loss of increment sensitivity on a bright background would be less than the loss of absolute cone sensitivity. ^{36 37 38 39} Studies using threshold-versus-luminance functions and/or dark adaptation functions would be helpful to quantify the relationship between rod and cone dysfunction in AMD. 

How do our results compare with those of previous studies of visual function in AMD patients at early disease stages? Most work has been performed under test conditions that mainly depend on cones. For example, there have been findings of impairment in color discrimination, ^{40 41} color matching, ^{11 42} flicker sensitivity, ⁴³ spatial contrast sensitivity, ^{41 44 45} low contrast acuity, ⁴⁴ photopic light sensitivity, ⁴¹ and focal cone electroretinogram parameters. ^{46 47} Losses in visual function under dark-adapted or low luminance conditions also have been reported, including impaired sensitivity in the fovea ^{8 9 11 14} and central visual field, ^{5 7 12 13} deficits in letter acuity, ¹⁴ and abnormalities in both rod and cone dark adaptation kinetics. ^{6 10 13 41 48 49} Studies using methods such as those in the current work showed dark-adapted perimetric abnormalities at some loci in at least half of 12 to 14 patients examined. ¹² In studies that explored global and peripheral retinal function with electrophysiological as well as psychophysical techniques, dark-adapted visual sensitivity losses in the central 20° were also noted in some patients. ^{5 50}  

The present study pointed to the existence of spatially extended dark-adapted sensitivity loss beyond the anatomic or even clinically defined macula and a regional pattern. This impairment tended to peak in the parafoveal region (2–4° or 1 mm eccentric) and decreased at increasing eccentricities. This observation is concordant with histopathology of donor retinas. After an area of peak rod loss between 0.5 and 3 mm eccentricity on the retina, rod loss falls off at further eccentricities. ⁴ The basis of such a gradient of vulnerability of rod system dysfunction across the central retina of AMD patients is not known. 

Reduced foveal absolute sensitivity to a long wavelength stimulus has been found in patients with high-risk drusen and been shown to be a predictor of advanced AMD. ^{8 9} We also found reduced dark-adapted sensitivity at the fovea (with a 500-nm stimulus); however, parafoveal results showed a significantly greater degree of dysfunction. This may not be unexpected considering that relative foveal sparing and parafoveal vulnerability have been noted in relation to macular degeneration, ⁵¹ and especially geographic atrophy in AMD. ⁵² We suggest that parafoveal dark-adapted impairment may be another useful early functional marker for patients whose fate is to progress to later stages of AMD. Future prospective studies would be necessary to substantiate this notion. 

Relationships between the degree of foveal dysfunction and the presence of macular drusen have been reported. ⁶ Stimuli placed on drusen under fundus visualization have shown that function was similar on and off the drusen. ⁷ Among our patients with early AMD, there was no correlation between amount of drusen and sensitivities measured foveally, parafoveally, dark-adapted or light-adapted. Thus, in our sample of patients, more large drusen did not equate with more dysfunction, suggesting that these particular fundus features and function tests are measuring different expressions of the AMD disease process. It is conceivable that other methods to detect fundus alterations, such as fluorescein angiography, indocyanine green angiography, infrared imaging, or fundus autofluorescence, ^{12 53 54} may have revealed changes that would better correlate with visual dysfunction. It also possible that patients with considerable large drusen and no dark-adapted impairment could show visual dysfunction by methods we did not use in this study. For example, some AMD patients with no measurable dark-adapted sensitivity loss have been shown to have abnormal kinetics of dark adaptation. ^{13 55 56} These findings and those from Mendelian genetic models of AMD with extensive sub-RPE deposits ^{57 58 59} suggest that rod dark adaptation kinetics can be perturbed without loss of rod system sensitivity. 

It is established that fellow eyes with drusen in patients with unilateral exudative AMD are at high risk to develop choroidal neovascularization, ^{60 61 62} and there can be significant abnormalities of foveal function despite good visual acuity. ⁶ We extend these observations by our report of increased dark-adapted sensitivity loss in the test eyes of patients with fellow eyes having unilateral disciform scars; these results are consistent with a report of night vision complaints being greater in such patients when compared with those having bilateral large drusen. ⁶³  

AMD is acknowledged to be a complex set of multifactorial diseases and the initial site of disease expression could be in the retinal pigment epithelium, Bruch’s membrane, photoreceptors, or other neighboring structures. ^{1 2 3} The recently discovered genetic causes of Mendelian inherited early-onset maculopathies that show some resemblances to AMD (e.g., RDS/peripherin, ⁶⁴ TIMP3, ⁵⁷ VMD2, ⁶⁵ ABCR, ⁶⁶ and EFEMP1 ⁶⁷ ) illustrate the diverse sites of initial molecular abnormalities that can lead to a macular degeneration phenotype. Whatever the initiating event(s) and exact pathogenic sequence in AMD, rod photoreceptors definitely can show dysfunction (and degeneration, ⁴ ) relatively early in the disease. From a practical viewpoint, parafoveal dark-adapted sensitivity levels in some AMD patients may thus be able to be exploited as a novel means to monitor disease progression or as an outcome measure of treatment efficacy to complement traditional foveal cone-based psychophysical measures. 

 Supported by National Institutes of Health (Grants R01-AG04212, R01-EY05627, T32-EY07033); The Foundation Fighting Blindness, Inc.; the Alabama Eye Institute; Research to Prevent Blindness, Inc.; F. M. Kirby Foundation; the Macular Disease Foundation; and the Mackall Trust.

 Submitted for publication June 22, 1999; revised August 27, 1999; accepted September 17, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Retinal locations of psychophysical test positions. Small circles (○) superimposed on a 45° fundus photograph indicate the size and the location of the 51 points tested. Black concentric circles are drawn for reference at eccentricities of 3°, 5°, 7°, 9°, 11°, and 13°.

Figure 2.

 

Frequency histograms of mean field dark-adapted (A) and light-adapted (B) sensitivity in AMD patients. Sensitivity is in decibel (dB) units. Dark-adapted sensitivities have been corrected for preretinal absorption. Box plots above each histogram compare the distributions of the AMD patients to a group of age-matched normal subjects. Box represents 25th and 75th percentile; whiskers, 10th and 90th percentile; solid line within the box, the median; and •, the mean. (A, B) Vertical line in the histogram represents the lower normal limit (mean− 2 SD).

Figure 3.

 

Selective dark-adapted sensitivity loss in AMD. (A) Mean versus SD in a group of 12 AMD patients (♦) with reduced mean field dark-adapted sensitivity but normal light-adapted sensitivity and in normal subjects (⋄). (B through E) Letters near symbols specify patients shown in the respective panels. Gray scale maps of dark-adapted sensitivity loss in 4 of the 12 AMD patients: two with relatively homogeneous dysfunction (B, C) and two with more pronounced dysfunction in a region (D, E). Data are displayed as right eyes; darker shades of gray indicate greater impairment (30 dB scale of losses, bottom right); ×, physiological blind spot.

Figure 4.

 

Regional variations in dark- (A) and light-adapted (B) sensitivity. Sensitivity loss as a function of eccentricity in all AMD patients (•). A subset of AMD patients (n = 36, ▴) are patients with normal dark- and light-adapted mean field sensitivities. ⋄, normal controls; F, fovea. Bars, ±1 SEM.

The authors thank R. Feist, M. White, Jr., G. Hammack, A. Augsburger, S. Orlin, and M. Sulewski for patient referrals; T. Aleman, M. Sharma, G. McGwin and J. Huang for data analyses; and G. Regunath, D. Hanna, L. Gardner, K. Mejia, and J. Christopher for help with data collection.