Subjects With Unilateral Neovascular AMD Have Bilateral Delays in Rod-Mediated Phototransduction Activation Kinetics and in Dark Adaptation Recovery

Abstract

Purpose.: To assess the impact of early (dry) and late (wet/neovascular and/or atrophic) forms of AMD on panretinal function.

Methods.: Light- and dark-adapted full-field ERG recordings were obtained over a 5-log-unit intensity range from both eyes of 25 patients with unilateral wet AMD. Fellow eyes showed various signs of dry AMD ranging from multiple medium-sized drusen to noncentral geographic atrophy. The leading edges of rod-isolated ERG a-waves were fitted to a quantitative model of phototransduction. ERG a- and b-wave amplitudes and implicit times were compared between wet and dry AMD eyes and from non-AMD eyes of age-matched subjects. A quantitative and objective assessment of dark adaptation was achieved by recording the recovery of the pure rod b-wave (postsynaptic depolarization of rod bipolar cells); b-wave amplitudes were measured at 120-second intervals for 20 minutes and normalized to the amplitude recorded at t = 20 minutes.

Results.: Delays in mixed a- and b-wave implicit times were recorded in both wet and dry AMD eyes. Time required to reach 50% of fully recovered responses was delayed in all wet and dry AMD eyes independently of dry AMD severity in the fellow eye. Generalized cone dysfunction and slower activation of the rod phototransduction cascade was noted in a subgroup of patients with advanced features of dry AMD in the fellow eye.

Conclusions.: Patients with unilateral wet AMD display rod dysfunction in both their wet and dry AMD eyes. A subset of these patients display, in addition, bilateral cone dysfunction and delayed rod phototransduction activation, which may either reflect extensive morphologic change in advanced stages of AMD and/or represent a distinct phenotypic manifestation within the heterogeneous context of AMD as a disease.

Introduction

AMD is the leading cause of blindness in the elderly of the Western world. ¹ Since the prevalence of AMD is expected to increase from 1.75 million individuals to 2.95 million individuals by 2020, ² it is essential to delineate the pathophysiological events that take place prior to vision loss in AMD. Early-stage AMD is characterized by changes in the Bruch's membrane, choriocapillaris and RPE that gradually impair photoreceptor viability. Late-stage AMD, on the other hand, is characterized by complete loss of the supporting RPE layer (geographic atrophy, most advanced form of dry AMD) in the macular region and/or abnormal growth of choroidal blood vessels (neovascular AMD, also called wet AMD). Although representing only 10% of all AMD cases, these advanced forms are responsible for 90% of the vision loss attributed to AMD. ³  

The risk factors involved in progression to advanced AMD include age, smoking history, and diet (more specifically antioxidant and omega-3 fatty acid intake). ⁴ In addition, several genetic determinants have been identified. ⁵ These include polymorphisms in genes involved in regulation of complement activity, lipid metabolism, extracellular matrix remodeling, and angiogenesis. ⁶  

The role of the alternate complement pathway appears crucial to disease pathogenesis and progression. ⁷ Activated complement components have been reported to correlate with disease stage in both the vitreous and the serum of patients with AMD. ^8,9 Together with previous evidence from psychophysical tests, ^10–12 these findings strongly support the fact that AMD is a systemic disease manifesting locally in the aging macula. ⁸ We can, therefore, hypothesize that the pathological and functional alterations should extend beyond the macula, and involve the whole retina. 

Numerous studies have relied on full-field ERG as a mean to assess functional changes across the whole retina. The ERG represents a mass response from the neural retina elicited by flashes of various intensities and wavelengths, under specific light adaptation levels. By controlling these variables, it is possible to isolate rod- from cone-driven contribution, as well as to distinguish pre- from postsynaptic activity of the mass potential response. Historically, due to the regional macular dysfunction typifying AMD, full-field ERG has not been used in the study of AMD progression. Since the macula represents only 5% of the whole retinal surface area, dysfunction confined to this region does not result in impaired full-field ERG response. ¹³ The multifocal ERG has been developed specifically to assess regional function, such as from the macula, but it is limited to cone-driven responses. Past full-field ERG studies have shown conflicting findings in AMD, most likely reflecting the complex nature of the disease. These findings include a global reduction in retinal function ¹⁴ and a generalized cone dysfunction in subgroups of AMD patients. ¹⁵ An increasing number of studies point to an early-onset rod dysfunction that precedes cone dysfunction in AMD. ^10,11,16 Jackson et al. ¹⁷ extensively studied the phototransduction activation and deactivation kinetics of rods as a function of AMD progression. Although neither the activation and deactivation kinetics were affected in dry AMD, ^17,18 the wet form was characterized by a dramatic delay of rod inactivation kinetics. ¹⁸  

The purpose of our study was to investigate panretinal rod and cone function in both eyes of individuals with already established neovascular AMD in a single eye. The rationale for choosing this population was that wet AMD in one eye represents the most reliable risk factor for the fellow eye (dry AMD) to progress to the neovascular form. ¹⁹ We hypothesized that ERG recordings from these high-risk individuals would reveal dysfunctions involving the whole retina in both eyes, therefore challenging the widely accepted notion of a disease confined to the macula. 

Materials and Methods

This study received ethics approval from the Health Research Ethics Board (Biomedical Panel) of the University of Alberta. All procedures conformed to the Code of Ethics of the World Medical Association (Declaration of Helsinki). ERG recordings were done with the understanding and written consent of each subject after detailed explanation of the nature and possible consequences of the study. All data were analyzed in an aggregated fashion. 

Patients

Patients from the Alberta Retina Clinic (ARC) were invited to participate in this study. Twenty-five subjects (mean age: 73.6, SD: 8.9) with unilateral neovascular macular degeneration (Age-Related Eye Disease Study [AREDS] severity scale ²⁰ : 11b; definite end-stage disease) were recruited in the study. Absence of wet AMD in the fellow eye was confirmed by OCT. All subjects were undergoing treatment in their neovascular eye with intravitreal injections of anti-VEGF agents. The number of injections did not differ among subjects (mean: 11 monthly injections). Subjects with cataracts or previous ocular surgery (with the exception of cataract surgery) were excluded. Other exclusion criteria included diagnosis of diabetic retinopathy, glaucoma, ocular hypertension or optic neuropathy, and any history of medication known to be toxic to the retina or optic nerve (deferoxamine, chloroquine, tamoxifen, chlorpromazine, phenothiazines, chronic systemic steroid use of more than 10 mg per day, and ethambutol). Eleven of the AMD subjects had undergone bilateral cataract surgery with intraocular lens placement. Data from a group of 18 age-matched (69 ± 5) controls were used for comparison. Twelve additional pseudophakic control subjects (71 ± 4) were used for the dark-adaptation analysis. Control subjects had best-corrected visual acuity 20/20 or better, with no current or past ocular history and their fundus appearance was free of drusen or pigmentary anomalies (AREDS scale ²⁰ : step 1). 

Fellow Eye Dry AMD Severity Grading

Modified three-field 30° stereoscopic color fundus photographs were acquired after dilation using a Visucam NM Pro (Visucam, Zeiss, Germany) fundus camera. Non-neovascular fellow eyes were evaluated for dry AMD severity based on the 9-step AREDS severity scale, ²⁰ which combines a six-step drusen area scale with a five-step pigmentary abnormality scale. All grading was performed by an independent former Wisconsin Reading Center grader (Jane Armstrong), masked to patient characteristics and ERG recording. For grouping and comparison purposes, eyes with AREDS scale scores of 1 to 3 were characterized as early AMD (AMD1), 4 to 6 as intermediate (AMD2), and 7 to 9 as advanced (AMD3). Subjects in AMD3 group had extensive areas of decreased pigment up to 0.5 DA (disc area) and/or noncentral geographic atrophy. Representative examples for each group are provided in Figure 1. 

Figure 1

 

Representative fundus photographs from the fellow eyes of patients with unilateral neovascular AMD. Top row: Eyes classified as AMD1 had several hard drusen or a few intermediate size drusen. Middle row: Eyes classified as AMD2 had large, soft drusen or several intermediate size drusen with increased pigmentation. Bottom row: Eyes classified as AMD3 had large drusen with extensive areas of depigmentation up to 0.5DA. Noncentral geographic atrophy was included in this category.

ERG Recordings

The subjects underwent bilateral full-field ERG recordings (Espion E ² system; Diagnosys LLC, Lowell, MA) in accordance with International Society for Clinical Electrophysiology of Vision (ISCEV) standard protocols ²¹ and supplemented by custom-made analytical protocols described below. Pupils were dilated using two drops of 1% Tropicamide applied on the corneal surface of each eye. Testing lasted 1.5 hours and began after full mydriasis was achieved (>6.5 mm pupil diameter); full mydriatic effect of the eye drops was confirmed in all subjects after the end of the testing period. Recording electrodes consisted of Dawson-Trick-Litzkow type (DTL) fiber electrodes with references electrodes (gold surface electrode, F-E5GH-60; Grass Technologies, West Warwick, RI) placed on the lateral edge of the orbital bone. The ground electrode (gold surface electrode) was centered on the forehead, 1 inch above the upper eyebrow line. All stimuli consisted of white flashes generated by a xenon bulb (6500°K color temperature, 10 μs duration) unless specifically noted. Calibration of light levels was achieved using an IL1700 photometer (International Light Technologies, Inc., Peabody, MA) equipped with either a photopic or scotopic filter. ²²  

Amplitudes and implicit times of the a- and b-waves were measured by strictly adhering to the standards provided by the ISCEV. ²¹ B/A wave ratios were examined only when both a- and b-waves exceeded criterion response (10 μV). Criterion amplitudes were set at 10 μV, a magnitude below which the signal could not be clearly distinguished from background noise. For all traces obtained, both a- and b-waves were clearly distinguishable when exceeding 10 μV in amplitude; both their amplitude and implicit times could be measured accurately. 

Photopic ERG

Pure cone-driven responses were studied under photopic conditions (30 cd/m², measured at the corneal surface), by presenting single white flashes (as described above) with increasing time-integrated luminances along 11 stepwise increments: −1.63, −1.22, −0.81, −0.42, −0.02, 0.38, 0.88, 1.37, 1.89, 2.39, and 2.86 photopic log cd s/m² (logarithm of candela seconds/meters square). Time interval between each step was 10 seconds and each stimulus was presented two to six times at 5-second intervals (six times for steps 1–8, two times for steps 9–11). Responses to flashes of a given strength were averaged at each step. Amplitudes and implicit times of the a- and b-waves were measured as described above. 

Dark Adaptation

Subjects were dark adapted for 20 minutes after conclusion of the photopic ERG recordings. Bleaching conditions were identical in all study subjects as previously published ²³ and consisted of a 10-minute exposure to 30 cd/m² background in addition to the last six sets of bright flashes (steps 9–11) of an average of 2.38 photopic log cd s/m² each. Consistency of this bleaching regimen among groups was confirmed by the observation that normalized pure rod b-wave amplitudes recorded at time zero (just after transition from light to dark-adapted background) were 5% to 10% of the fully dark-adapted values. To study the process of dark adaptation, the amplitude of the pure rod b-wave was assessed every 120 seconds (starting at t = 0 minutes, 11 steps in total) using the standard ISCEV dim flash of −2.04 log cd s/m². Each step consisted of five repeats with 5-second intervals between them. Responses were averaged at each step. 

Dark-Adapted ERG

Dark-adapted subjects were stimulated using 16 flashes of increasing strengths (−5.22, −4.70, −4.22, −3.70, −3.22, −2.82, −2.44, −2.04, −1.63, −1.22, −0.81, −0.42, −0.02, 0.38, 0.88, and 1.37 scotopic log cd s/m²), each averaged three to five times. To allow for maximal rod recovery between consecutive flashes, interstimuli intervals were increased from 4 seconds at the lowest stimulus strength up to 60 seconds at the highest stimulus strength. ²⁴  

To isolate rod photoresponses, photopic ERG responses were subtracted from the mixed dark-adapted responses elicited by stimuli of the same intensities; this exercise was applied for the responses elicited by the three highest intensity stimuli (0.38, 0.88, and 1.37 scotopic log cd s/m²). For each subject, these isolated rod responses were fitted to the following equation to describe the response (R) as function of flash intensity (I), and time (t), based on the Hood and Birch formulation ²⁵ of the Lamb and Pugh model ²⁶ :  

 

where S is sensitivity, t_d is the delay before onset of the a-wave, and Rm_{P 3} is the maximum amplitude (Fig. 2). All curve fits were performed using Igor Pro (Wavemetrics, Inc., Lake Oswego, OR) and satisfied a least-squares goodness-of-fit criterion (R ²) of at least 0.90. Best-fit variables S and Rm_{P 3} were calculated and transformed to log values for data analysis.  

Figure 2

 

Fitting of rod-isolated ERG a-wave leading edge. Representative ERG recordings (solid curves) and best fit (dashed curves) from an age-matched non-AMD eye (A), dry AMD eye (B), and wet AMD eye (C) (same patient for [B, C]) elicited by white flash stimuli ranging in energy from 0.93 to 3.36 log phot td-s; note that a tight fit was obtained independently of the stimulus intensity and state of the eye (non-, dry, or wet AMD).

Statistics

Statistical significance between groups was assessed using repeated-measures ANOVA with the Greenhouse-Geisser correction for sphericity. Post hoc analyses were done between the individual groups and at individual stimulus strengths or time points using the Bonferroni technique for multiple comparisons. Graphpad Prism was used for one-way ANOVA, linear regression, and correlation analyses (Graphpad Software, Inc., La Jolla CA). SPSS was used for the repeated-measures ANOVA (SPSS, Inc., Chicago, IL). Significance was set at P less than 0.05. Data points on graphs represent mean ± 95% confidence interval (95% CI). 

Results

Light-Adapted Responses

Figures 3A through 3E provide the intensity-response curves for the photopic a- and b-waves. There were no differences in any of the related parameters (amplitude, implicit time, and b/a ratio) among groups. For instance, in all subjects, the b-wave luminance response function followed the pattern of the previously described “photopic hill” ²⁷ : initial amplitude increase as a function of flash strength increments, peaking halfway along the x-axis, followed by a nadir (this pattern was named due to its resemblance to a “hill”). The peak of this “photopic hill” always occurred at the same light strength independently of the group. 

Figure 3

 

Pure cone-driven ERG responses elicited over a photopic background with stimuli ranging in intensity from −0.02 to 2.86 photopic log cd s/m². Intensity responses for a-wave amplitude (A) and implicit time (B) as well as for b-wave amplitude (C) and implicit time (D), and finally for b/a ratio (E) are provided for the three eye groups studied: controls (black), dry AMD (blue), and wet AMD (red). (F) Maximal b-wave amplitudes were calculated for each eye per group; dry AMD eyes were divided into three severity levels: AMD1, AMD2, and AMD3.

The maximal b-wave amplitude (b_max) was recorded with a flash intensity of 7.67 cd s/m² photopic units. This intensity was selected as it preceded the onset of the photopic hill in all subjects. To test whether maximal b-wave amplitude was related to macular phenotype severity, a subgroup analysis was performed. Scatter plots of cone b-wave amplitudes in all subgroups are illustrated in Figure 3F. Even though no overall group mean difference was reported (P = 0.15), 19 of 45 eyes had values below the 2.5% percentile of our control data set. Subjects with AMD3 features in the fellow eye (displaying morphological features of nonexudative AMD and extensive areas of hypopigmentation or noncentral geographic atrophy) were more likely to fall into this category. 

Dark Adaptation

During the 20-minute dark adaptation, there was a progressive increase between five- and six-fold in pure rod b-wave amplitude for all groups (Fig. 4). Due to significant interindividual variability in fully recovered amplitudes at t = 20 (coefficient of variation > 30%), normalized values at each time point were used for all subjects (Fig. 4). Averaged data points were fitted to a sigmoid function of the form:  

 

where y represents normalized b-wave amplitude; base represents normalized pure rod b-wave amplitude values at the beginning of the dark adaptation (t = 0); thalf represents the time to reach half the maximum normalized amplitude value (0.5); t represents time, and t ₁₀ represents the inverse slope as measured at 10 minutes (at which point linearity of the sigmoid curve was optimal). This inverse slope has the dimensions of time. All curves met a goodness of fit (R ²) of at least 0.95. Best-fit parameters for thalf and rate were calculated for all groups (Fig. 5). Base values were identical for all three groups (0.04 ± 0.02 for non-AMD eyes, 0.04 ± 0.02 for dry AMD eyes, and 0.04 ± 0.02 for neovascular AMD eyes). Therefore, the marginal rod contribution prior to beginning dark adaptation was not affected by AMD status. Best-fit coefficient values for thalf, the time to reach half of the fully recovered pure rod b-wave amplitude (obtained at the end of the dark adaptation, i.e., t = 20 minutes), were significantly higher (P < 0.0001) in AMD eyes. These were estimated to be 9 ± 0.4 minimum for the non-AMD group, 13 ± 0.4 for the dry AMD eyes, and 13 ± 0.3 for the wet AMD eyes. In addition, the slope parameter of the fitted curve (rate) was reduced compared to controls (P = 0.0008), implying a delay in recovery (not illustrated). The dynamic of recovery was equally delayed in both neovascular (wet AMD) and their fellow eyes (dry AMD). Inclusion of all subjects or subanalysis limited to pseudophakic subjects yielded similar results. As such, to eliminate potential contributions of various levels of lens opacification, we restricted data presentation to subjects with artificial lenses in both eyes (n = 12 AMD and n = 11 control subjects).  

Figure 4

 

Dark adaptation results: raw data. To ensure optimal clarity of the individual data points (circles), data from both eyes of all respective AMD subjects were pooled (hollow black circles); the rationale for this grouping was to distinguish values of control eyes from those of AMD eyes. The y-axis represents the pure rod b-wave amplitude normalized against amplitude at 20 minutes of dark adaptation; x-axis represents time in minutes following transition from light to dark adaptation. Fitted lines (all dashed): controls (black), dry AMD (blue), and wet AMD (red).

Figure 5

 

Dark adaptation results: time to reach 50% recovery of pure rod b-wave amplitude (A) and amplitude reached at 20 minutes of dark adaptation (B). Asterisk indicates that there was a statically significant difference for time to reach 50% recovery, between control and dry AMD eyes as well as between controls and wet AMD eyes (P < 0.05). Amplitudes of the pure rod b-wave recorded at 20 minutes of dark adaptation were not different among the three eye groups. Controls (black), dry AMD (blue), and wet AMD (red). Error bars represent SD.

Dark-Adapted Responses

Figures 6A through 6E provide the intensity-response curves for the dark-adapted mixed a- and b-waves. Neither a- or b-wave response curves showed any significant differences among the three groups with regard to amplitude (repeated measures ANOVA P values 0.65 and 0.31, respectively). Within the range of stimulus intensity tested, the mean maximal recorded mixed a-wave amplitude was −211 ± 45 in non-AMD eyes versus 211 ± 52 in nonneovascular fellow eyes and 207 ± 45 in neovascular eyes. Mean maximal b-wave values were 329 ± 76, 319 ± 70, and 311 ± 70, respectively. There was no difference noted between neovascular and fellow eyes (Wilcoxon matched-pairs signed rank test: P = 0.20 for maximal a-wave and P = 0.38 for maximal b-wave). 

Figure 6

 

Mixed rod-cone–driven ERG responses elicited following 20 minutes of dark adaptation under a scotopic background with 16 flashes of increasing strengths encompassing −5.22 to 1.37 scotopic log cd s/m². Intensity responses for a-wave amplitude (A) and implicit time (B) as well as for b-wave amplitude (C) and implicit time (D), and finally for b/a ratio (E) are provided for the three eye groups studied: controls (black), dry AMD (blue), and wet AMD (red). (F) Averages (±95% confidence intervals, dashed lines) of normalized mixed a-waves elicited by the highest intensity stimulus (1.37 scotopic log cd s/m²). The asterisk denotes a significant shift in implicit time (“shift to the right” as pointed by the arrow) between control eyes and eyes with either dry or wet AMD (P < 0.05). Controls (black), dry AMD (blue), and wet AMD (red).

Implicit times were significantly delayed for both a- and b-waves throughout all stimulus intensities tested. Delays were more pronounced for a-waves (P < 0.0001). For more accurate determination of a-wave parameters, implicit time measurement was carried out at the highest available intensity (1.37 photopic log cd s/m²). Such high-intensity stimuli are now recommended by ISCEV in routine testing since they yield a larger a-wave with improved definition and lack of double trough. ²¹ Mean implicit time of the maximal mixed a-wave was 14 ± 1 in non-AMD eyes, 16 ± 1 in dry AMD fellow eyes, and 16 ± 1 in wet AMD eyes (analysis of variance: F statistic 33.19, P < 0.0001). 

B-wave implicit times were also significantly different among the three groups of eyes for all intensities tested (repeated measures ANOVA: P = 0.02). The mean implicit time of b-waves at their maximal amplitude was equally delayed in both eyes of AMD subjects when compared to non-AMD eyes (52 ± 6 for non-AMD eyes compared to 57 ± 7 for the neovascular and 58 ± 7 for the non-neovascular fellow eyes; analysis of variance: F statistic 8.4, P < 0.004). 

To further investigate the origin of the delays in a-wave implicit times in AMD versus non-AMD eyes, normalization of the a-wave function was performed for each response elicited at the highest stimulus intensity. Such an approach allows controlling for interindividual differences in a-wave amplitude. Normalized photoresponses provide an estimate of the time course of cGMP-channel closure ²⁸ through the following equation:  

 

where a_max represents the maximal a-wave amplitude and F(t) represents the function of cGMP-activated current as a fraction of resting (dark) level. Group averages with 95% CIs (dashed lines) of normalized photoresponses are provided in Figure 6F. Results from fitting of the a-wave leading edge indicate that phototransduction activation kinetics were significantly delayed in both eyes of AMD subjects (to a similar extent) when compared to non-AMD eyes.  

Phototransduction Activation Kinetics in Rods

To address whether rod-mediated activity accounts for these observed delays in AMD versus age-matched non-AMD eyes, parameters of activation of the rod photoresponse (Rm _P3, S, and t_d ) were estimated using the Hood and Birch modification of the Lamb and Pugh model (see Methods section). The Table summarizes all estimated rod-mediated ERG parameters. Maximum amplitude of the rod photoresponse (Rm _P3) did not differ among the three groups of eyes (P = 0.9580), whereas sensitivity (S) log values and time delay (t_d ) showed significant differences. Mean logS was estimated to be 0.98 ± 0.19 in non-AMD eyes, 0.79 ± 0.18 in neovascular eyes, and 0.78 ± 0.26 in nonneovascular eyes, yielding differences between AMD and non-AMD (P = 0.0039). Significance remained when comparison was limited to AMD subjects with intraocular lenses (P = 0.005). The time delay (t_d ) parameter was allowed to vary in the ERG response model rather than kept constant. On average it was estimated to be approximately 0.6 to 0.7 seconds higher in some AMD subjects (3.84 ± 0.53 vs. 4.47 ± 0.69) (see section below). 

Table

 

Rod-Mediated ERG Parameters

	Age-Matched Controls, n = 25	Dry AMD, n = 20	Wet AMD, n = 18	P Value
logS	0.92 (0.19)	0.78 (0.26)	0.79 (0.18)	0.0039
logS/IOL	0.98 (0.14)	0.78 (0.16)	0.79 (0.17)	0.0072
log Rm P3	2.27 (0.08)	2.28 (0.12)	2.27 (0.10)	0.9580
td	3.84 (0.53)	4.47 (0.69)	4.37 (0.61)	0.002

 Values in brackets represent SD.

Fellow Eye Macular Phenotype and Rod-Mediated ERG Parameters

To investigate whether maculopathy severity level in the non-neovascular fellow eye was related to any of the rod-mediated ERG parameters, a subgroup analysis of variance was performed (n = 6 eyes AMD1; n = 8 AMD2, and n = 6 in AMD3 group; see Methods section). Subjects in the AMD3 group tended to have extensive areas of decreased pigment up to 0.5 DA and/or noncentral geographic atrophy. Scatter plots of logS, log Rm_{P 3}, and t_d values are illustrated in Figure 7 for all groups. In terms of sensitivity (S), post hoc analysis with Bonferroni correction for multiple comparisons, yielded, interestingly, significance only for the AMD3 subgroup for both their non-neovascular and neovascular eyes (P = 0.002 for both). Rm_{P 3} did not vary significantly among subgroups (P = 0.67). Time delay (td) was the most variable ERG parameter and remained significant only in the AMD2 group (P = 0.029) when correction for multiple comparisons was applied. 

Figure 7

 

Phototransduction activation parameters obtained from fitting the leading edge of the isolated rod a-wave: (A) log sensitivity (logS); (B) maximal amplitude (log Rm _P3); (C) delay before onset (t_d ). Asterisk indicates statistically significant differences between groups.

Discussion

Our full-field ERG results indicate that a subset of AMD patients undergo dysfunction affecting both rod- and cone-driven retinal processing that span beyond the macula, across the whole retina. For instance, delayed recovery of the pure rod b-wave following partial rod bleaching was noted in both eyes of patients with unilateral wet AMD. These results provide the first objective evidence of a whole retina-based impairment of dark adaptation. Abnormalities of dark adaptation in AMD have been documented primarily through psychophysical studies in the past. ²⁹ Various groups have explored the validity of relying on the pure rod b-wave amplitude as an objective complement to psychophysical dark adaptation tests with success. Adrian ³⁰ first showed that the b-wave was reduced substantially after large bleaches and that it gradually recovered with time. Karpe and Tansley ³¹ and Fulton and Rushton (1978) ³² also reported that the time course of b-wave amplitude recovery corresponds closely to the recovery of the visual threshold, measured psychophysically. Rod outer segments contain the photosensitive visual pigment known as rhodopsin, which consists of a visual chromophore (11-cis retinal) covalently bound to a rod specific protein, opsin. Absorption of photons by pigments results in photoisomerization of the chromophore to all-trans-retinal and generation of free opsin. Recycling of retinal to its 11-cis form and regeneration of rhodopsin is achieved through a well-characterized retinal pigment epithelium (RPE) pathway called retinoid visual cycle. ^33,34 Despite differences in the psychophysical and ERG experiments, two consecutive works by Cameron et al. ³⁵ and Ruseckaite et al. ³⁶ showed that both methods extract a common time course for the main component of rod sensitivity recovery (commonly referred to as “S2” component ³³ ). Lamb and Pugh ³³ reviewed a wide range of evidence indicating that this component represents in molecular terms “unregenerated opsin.” Our results suggest that the time course of rod b-wave amplitude recovery is delayed in both eyes of patients with unilateral neovascular (wet) AMD. Reasons underlying this defect remain to be explored. Since the component measured to assess dark adaptation was the pure rod b-wave, changes directly affecting its generation (depolarization of ON type bipolar cells by rods) must be examined. 

Saari et al. ³⁷ suggested that all-trans retinal released by bleaching may have an effect itself in reducing rod bipolar cell sensitivity. It is known that large bleaches lead to the formation of millimolar levels of all-trans retinal in the outer segments, ³⁷ and that all-trans retinal leads to closure of cyclic nucleotide-gated ion channels of photoreceptor outer segments ^38–40 and blocks the calcium channels of the synaptic terminal. ⁴¹ Either of these mechanisms would be expected to be associated with desensitization of the rod bipolar cell response. 

Another possibility might be related to the loss of DHA that has been suggested to occur in AMD. ⁴² Lower retina DHA levels have been associated with decreased binding affinity of IRBP (interphotoreceptor retinoid binding protein). ⁴³ IRBP is a key transport protein involved in photopigment regeneration, and contains high-affinity DHA-binding sites for chromophores. ⁴⁴ Alterations in retinal lipid content may also impede the efficiency of the RPE visual cycle and therefore account, at least in part, for the observed delays in dark adaptation. Furthermore, other factors likely to delay the processing of visual cycle retinoids in AMD could be intrinsic abnormalities of the RPE cell layer itself. Several lines of evidence suggest that excessive accumulation of lipofuscin by RPE cells is significant in terms of the etiology of AMD. ⁴⁴ Delayed rod dark adaptation is a common feature in patients with Stargardt's disease, an inherited juvenile form of macular degeneration that is characterized by excessive accumulation of lipofuscin and shares several phenotypic similarities with the dry form of AMD. ⁴⁵ To further decipher the exact origin of the observed rod dark-adaptation delays, more experiments are undoubtedly required. A key aspect involves relying on a double-flash approach as a means to isolate the photoresponse (also called PIII component) and examine its contribution to the reported panretinal defects. Previous double-flash recordings in AMD subjects by Jackson et al. ¹⁸ show a significant delay in rod inactivation kinetics in late stages of AMD. We therefore must consider the possibility that the dark-adaptation delays (as measured by recording the pure rod b-wave) are caused, at least in part, by presynaptic events in photoreceptors and RPE. 

Our study also provides additional evidence for panretinal defects in AMD: implicit times of dark-adapted a- and b-waves were prolonged in both dry and wet AMD eyes. Occurrence of such defects in both types of AMD suggests an underlying panretinal dysfunction that precedes the onset of neovascularization. Mechanisms responsible for increased implicit times can be related to generalized ischemia or overt activation of inflammation affecting the health of the entire neural retina. It is widely known that fellow eyes of advanced AMD patients are at the highest risk of developing choroidal neovascularization (CNV). The role of ischemia in the development of CNV has been well established. Systematic decrease in choroidal circulatory parameters has been observed with an increase in the severity of AMD features associated with risk for the development of CNV. ⁴⁶ Pathological changes, such as drusen accumulation and thickening of Bruch's membrane, increase the distance between the choriocapillaris and the retina, reduce the oxygen flux to the photoreceptors, and induce hypoxia in these cells. ⁴⁷ It is plausible that the ERG delays measured from the entire retina might predict the onset of CNV development in the fellow eye. Analogously, increased cone b-wave implicit times in 30 Hz flicker electroretinography have been associated with ocular neovascularization in cases of central retinal vein occlusion. ^48,49 It would be relevant to prospectively follow this cohort of patients to address this hypothesis. Of interest, we examined whether there was an association between the number of injections (mean: 11 monthly injections) and the degree of panretinal dysfunction as assessed with the ERG, and found there was none among the 25 unilateral wet AMD subjects we studied. However, we cannot exclude that repeated intravitreal delivery of anti-VEGF agents may have impaired photoreceptor function, not only in eyes with wet AMD but also in fellows eyes via a systemic delivery. ⁵⁰  

Overt activation of inflammation could also account for the observed delays in implicit times. Systemic and vitreal activation of the alternative complement pathway has been associated with genetic variants of CFH. ^8,9 Certain CFH haplotypes are known to confer high risk for developing advanced forms of the disease. A possible genotype-based association with a distinct ERG phenotype may exist. Such a genetic association might even be stronger in the subset of patients who exhibit a complementary generalized cone dysfunction. Such a subgroup has been consistently described even in earlier forms of the disease by Ronan et al. ¹⁵ and Ladewig et al. ⁵¹ and not only restricted to subjects with advanced morphologic changes in the retina, as reported in our study. 

To investigate further whether implicit time delays in AMD are of receptoral or postreceptoral origin, we assessed rod-mediated parameters of phototransduction activation. Reduced sensitivity was observed only in those AMD patients with advanced nonexudative morphology in the fellow eye, such as extensive area of RPE depigmentation. Jackson et al. ¹⁷ studied rod-mediated responses during the early stages of the disease and found no abnormalities. Their results showed that among all macular phenotypic characteristics, increased retinal pigment had the least association with decreased sensitivity of activation. Several groups have reported a strong correlation between intermediate to advanced macular phenotypes and peripheral retinal involvement, as manifested through the presence of drusen and reticular pigment near the equator. Lewis et al. ⁵² first described the histopathology of peripheral reticular pigment and showed an association with macular degenerative abnormalities. Intriguingly, Seddon et al. ⁵³ associated peripheral retinal involvement with certain CFH genotypes (CFHrs1410996 variant and CFHY402H). Even though whole retina phenotype assessment and genotyping were not within the scope of our study, these reports provide further support that a distinct genotype may underlie certain ERG characteristics. 

Acknowledgments

The authors thank distinguished researcher Jane Armstrong, former grader at the University of Wisconsin Fundus Photograph Reading Center, for her assistance in grading fundus photographs, and Judd Payne for his invaluable contribution in patient recruitment and ERG recordings. 

Supported by Canadian Institutes of Health Research (CIHR MOP 79278); Alberta Innovation Health Services (AIHS) establishment Grant 200700584; Olive Young Special Purpose Fund for Ophthalmology Research; and The Lena McLaughlin AMD Research Fund. ISD is an Alexander S. Onassis Foundation scholar and received an AIHS graduate studentship. YS is an AIHS Senior Scholar. 

Disclosure: I.S. Dimopoulos, None; M. Tennant, None; A. Johnson, None; S. Fisher, None; P.R. Freund, None; Y. Sauvé, None