Retinal Sensitivity to Flicker Modulation: Reduced by Early Age-Related Maculopathy

Abstract

purpose. To evaluate retinal, cone-mediated flicker sensitivity (CFS) in age-related maculopathy (ARM) by quantifying response gain and threshold of the focal electroretinogram (FERG) to flicker modulation.

methods. Nineteen patients with ARM (visual acuity ≥20/30) and 11 age-matched control subjects were examined. Twelve patients had less than 20 soft drusen in the macular region and no hyper-/hypopigmentation (early lesion), whereas seven had more than 20 soft drusen and/or focal hyper-/ hypopigmentation (advanced lesion). Macular (18°) FERGs were elicited by a sinusoidally flickering (41 Hz) uniform field (on a light-adapting background) whose modulation depth was varied between 16.5% and 94%. Amplitude and phase of the response’s fundamental harmonic were measured.

results. In both control subjects and patients with ARM, log FERG amplitude increased with log stimulus modulation depth with a straight line (power law) relation. However, the slope (or gain) of the function was, on average, steeper in control subjects than in patients with either early or advanced lesions. Mean FERG threshold, estimated from the value of the log modulation depth that yielded a criterion response, did not differ between control subjects and patients with early lesions but was increased (0.35 log units) compared with control subjects in those with advanced lesions. In both patient groups, but not in control subjects, mean FERG phase tended to delay with decreasing stimulus modulation depth.

conclusions. Retinal CFS losses can be detected in ARM by evaluating the FERG as a function of flicker modulation depth. Reduced response gain and phase delays, with normal thresholds, are associated with early lesions. Increased response thresholds, in addition to gain and phase abnormalities, may reflect more advanced lesions. Evaluating CFS by FERG may directly document different stages of macular dysfunction in ARM.

Age-related maculopathy (ARM ¹ ) is a degenerative disorder of the macula involving the retinal pigment epithelium (RPE), Bruch’s membrane, and photoreceptors. Early ARM is characterized ¹ by the presence of large, soft drusen and hyperpigmentation and/or hypopigmentation of the RPE. Late ARM, the potentially blinding stage of the disease, includes geographic atrophy of the RPE (dry age-related macular degeneration), or subretinal neovascular membranes (wet age-related macular degeneration). Early ARM is associated with anatomic and functional impairment of both rod and cone photoreceptors. ^{2 3 4 5} Cone function abnormalities have been documented by various psychophysical tests (see, for example, References 6 through 9) including spatial contrast sensitivity, color vision, recovery time after bleaching, and temporal cone flicker sensitivity (CFS). The temporal CFS function ¹⁰ has received increasing clinical interest because sensitivity losses at several temporal frequencies (10–40 Hz) have been found to be associated with early fundus alterations (i.e., drusen and/or RPE atrophy), ⁸ and to discriminate between ARM-affected eyes that convert to the exudative form and nonconverting eyes. ^{8 11 12}  

Retinal CFS can be evaluated by using the focal electroretinogram (FERG), ^{13 14} a signal generated from the macular region in response to flicker stimulation. The FERG is thought to reflect the activity of cone photoreceptors and bipolar cells ^{15 16} and can provide an index of the function of the outer retina. ^{14 15 17} Electroretinographic flicker modulation thresholds can be estimated either by a swept stimulus lock-in retrieval method ^{13 14 15 17 18} or by conventional averaging, ^{19 20} with extrapolation to a criterion amplitude. The function relating FERG response (i.e., its fundamental harmonic component) amplitude to stimulus modulation depth, at constant mean luminance, can be well described ²⁰ by a straight-line relation on log–log coordinates (power law relation). A power law relation with an exponent (or slope, on log–log coordinates) equal to unity defines a linear response gain. ²¹ A slope greater or less than unity indicates nonlinear response gain. ^{20 21} In normal subjects, the FERG amplitude versus modulation depth function may show either nonlinear or linear gain, depending on the stimulus’s temporal frequency. ^{20 22} Indeed, a response amplitude compression (slope <1) and an amplitude expansion (slope >1) are observed ²⁰ at low (16 Hz) and medium-high (40–48 Hz) temporal frequencies, respectively. Approximately linear gain can be found only at high (≥56 Hz) temporal frequencies. ²⁰ Changes in the FERG response gain are thought to result from one or more nonlinear gain control mechanisms, acting at the level of the distal retina. ^{20 22} There is also evidence that the retinal gain control mechanism acts by adjusting the phase delay of the response. ²⁰  

Direct assessment of outer retinal sensitivity by using the FERG may help to evaluate early cone dysfunction in ARM. Histopathologic studies ³ have demonstrated cone photoreceptor degeneration, with shortened outer segments and broadened inner segments, in most eyes with early ARM lesions. The functional consequences, at retinal level, of these receptoral abnormalities have not been characterized yet. In addition, testing FERG modulation sensitivity is of potential interest for evaluating the progression of dysfunction and the effects of treatments intended to preserve and/or rescue cone photoreceptors. In the present cross-sectional study, FERGs as a function of sinusoidal flicker modulation depth were recorded in patients with early ARM and in age-matched control subjects. FERG response gain, threshold, and phase characteristics were quantified and statistically compared between groups. The primary goal was to evaluate whether, and to what extent, the response-versus-modulation depth function was altered in early ARM. As a secondary purpose, we sought to determine whether abnormalities in the response function’s parameters were associated with the severity of ARM lesions, assessed by clinical and photographic methods. The results in patients showed significantly altered FERG functions, with increasing losses associated with the severity of lesions. The data of this study have been presented in part in abstract form. ²³  

Methods

Subjects

Nineteen patients (mean age: 67 ± 7; range: 54–84 years; 7 men and 12 women) with a diagnosis of bilateral ARM were included in the study. Demographic and clinical data of patients are reported in Table 1 . Each patient underwent standard general and ophthalmic examination. Clinical diagnosis of ARM was established by direct and indirect ophthalmoscopy and retinal biomicroscopy, when any of the following primary lesions ¹ in the macular area (i.e., the area within an eccentricity of approximately two disc diameters from the fovea) was identified: soft, distinct or indistinct drusen; areas of hyperpigmentation associated with drusen; or areas of hypopigmentation of the RPE, without any visibility of choroidal vessels, associated with drusen. All patients met the following inclusion criteria: best corrected visual acuity of 20/30 or better in the study eye, central fixation (assessed by direct ophthalmoscopy), normal color vision with Farnsworth D-15 testing, no signs of other retinal or optic nerve disease, and clear optical media. None of the patients had systemic disease (e.g., diabetes, systemic hypertension) or was taking medication (e.g., chloroquine) known to affect macular function. The eye with the best visual acuity underwent the full FERG protocol. If both eyes had equal acuity, one eye was randomly selected. In all patients, clinical evaluations (including fundus grading, described later) and FERG testing were performed within 2 to 3 weeks of each other. 

Eleven healthy volunteers, whose age (mean: 65 ± 7; range: 54–84 years) and sex distribution (4 men and 7 women) were comparable to those of the patients, were also tested (see Table 1 ). None of the control subjects showed ARM lesions during fundus examination performed by direct ophthalmoscopy and retinal biomicroscopy. Normal subjects met the same inclusionary criteria as the patients, except for best corrected visual acuity, which was 20/20 in all cases. 

The research followed the tenets of the Declaration of Helsinki. Informed consent was obtained from each patient or control subject before his or her inclusion in the study and after the goals and procedures of the research were fully explained. 

Fundus Grading

ARM lesions of the study eyes were graded on stereoscopic fundus photographs of the central 30° of the posterior pole (centered on the fovea). A macular grading scale based on the international classification and grading system ¹ was used. A single grader evaluated the photographs while masked to subject characteristics and FERG results. The presence of drusen and focal RPE hypo- and/or hyperpigmentation were noted within each of the nine subfields delimited by the scoring grid (see also Reference 24). Drusen were graded for size, type, area, and confluence. Focal RPE hyper- or hypopigmentation was graded from none to 50% or more of the examined area. Seventeen of the 19 patients included in the study also underwent fluorescein angiography according to standard techniques. ²⁵ Although fluorescein angiography results were not used for quantitative fundus grading, they provided a more accurate qualitative analysis of macular lesions. For instance, fluorescein angiography confirmed the presence of an RPE defect, with or without drusen, through detection of pathologic hyperfluorescence (i.e., transmitted hyperfluorescence), and helped to exclude lesions more typical of late ARM, such as geographic atrophy or serous RPE detachment. According to the results of grading, all study eyes were diagnosed as having ARM and, according to the classification also used by Jackson et al., ⁴ as belonging to stage 2 (i.e., one or more large drusen ≥63 μm) and/or focal hyperpigmentation. Study eyes were further classified according to the total drusen number and the presence or absence of focal RPE abnormalities. Twelve eyes had less than 20 large drusen (average number: 10; range: 5–15) and no hypo- or hyperpigmentation. Seven eyes had more than 20 large drusen (average number: 25; range: 22–28) and/or focal RPE abnormalities extending into at least 20% of one of the middle subfield areas. For the purposes of this study, the former group will be labeled as early lesion (EL) and the latter as advanced lesion (AL). 

Electrophysiological Methods

The flicker stimulus for macular FERG was a circular uniform field (subtending 18° in diameter, 80 candelas[ cd]/m² mean luminance), whose luminance was modulated sinusoidally at 41 Hz, presented on the rear of a Ganzfeld bowl illuminated at the same mean luminance as the stimulus. The temporal frequency of 41 Hz was chosen because preliminary results of our laboratory and previous findings ²⁶ indicated that at approximately this temporal frequency the main FERG response component (i.e., the fundamental harmonic) displays its maximum amplitude. The flickering uniform field was generated by an array of eight red LEDs (maximum λ: 660 nm; half-height bandwidth: 35 nm) sinusoidally driven by a programmable function generator. ²⁷ A diffusing filter placed in front of the LED array made it appear as a circle of uniform red light. A steady DC signal maintained the mean luminance at 80 cd/m². Both modulation depth and mean luminance were digitally controlled with an accuracy of 0.05 log units. Linearity of the LED intensity output from the digital function generator was calibrated by using a photodetector. Sine waves as a function of modulation depth were acquired and Fourier analyzed with the same routine used for the FERG response analysis (see later description). The function relating photodetector output to modulation depth was linear within the resolution limits of the recording system. In the recording protocol, the stimulus field was presented at six different modulation depths—quantified by the Michelson luminance contrast formula: 100% · (L _max − L _min)/(L _max+ L _min), where L _max and L _min are maximum and minimum luminance, respectively—between 16.5% and 93.8%, in 0.1- to 0.3-log unit steps. The lower end of the modulation series (i.e., 16.5%) was chosen because control experiments in normal subjects indicated that this value yielded response amplitudes closest to, but still significantly higher than, noise level. 

FERGs were recorded monocularly by means of Ag-AgCl superficial cup electrodes taped over the skin of the lower eyelid. A similar electrode, placed over the eyelid of the contralateral, patched eye, was used as reference (interocular ERG). ²⁸ As the recording protocol was extensive, the use of skin electrodes with an interocular recording represented a good compromise between signal-to-noise (S/N) and signal stability. Discussion on the FERG by skin electrodes and its relationship with the responses obtained by corneal electrodes can be found elsewhere. ^{29 30 31} FERG signals were amplified (100,000-fold), band-pass filtered between 1 and 250 Hz (−6 dB/octave), sampled with 12-bit resolution, (2-kHz sampling rate), and averaged. A total of 1600 events (in eight blocks of 200 events each) were averaged for each stimulus condition. The sweep duration was kept equal to the stimulus period. Single sweeps exceeding a threshold voltage (25 μV) were rejected, to minimize noise coming from blinks or eye movements. A discrete Fourier analysis was performed off-line to isolate the FERG fundamental harmonic, whose amplitude (in μV) and phase (in degrees) were estimated. Component amplitude and phase were also calculated separately for partial blocks (200-event packets) of the total average, from which the SE of amplitude and phase estimates were derived to test response reliability. ³² Averaging and Fourier analysis were also performed on signals sampled asynchronously at 1.1 times the temporal frequency of the stimulus, to give an estimate of the background noise at the fundamental component. An additional noise estimate at the fundamental was obtained by recording responses to a blank, unmodulated field kept at the same mean luminance as the stimulus. In all records, the noise amplitudes recorded with both methods were 0.048 μV or less. 

In all subjects, the FERG testing protocol was started after a preadaptation period of 20 minutes to the stimulus mean illuminance, to avoid gradual changes in light adaptation during the experiment. ³³ Subjects’ pupils were pharmacologically dilated (1% tropicamide) to at least 8 mm, and no differences were detected between patients and control subjects (control subjects’ mean pupil diameter: 8.8 ± 0.4 mm; patients’ mean diameter: 8.9 ± 0.3 mm; t = 0.62, P not significant). Individual pupil diameters, measured during the FERG recording session, are reported in Table 1 . Subjects fixated (from a distance of 30 cm) at the center of the stimulation field with the aid of a small (15 minutes of arc) fixation mark. In each patient, eyelid opening and blink frequency during the recording session were judged to be normal by an observer who monitored patients’ fixation. A FERG response was first collected at the maximum modulation depth (93.8%) included in the protocol and was evaluated for reliability and S/N. In all control subjects and patients, the responses satisfied the following criteria: SD estimates of less than 20% (variation coefficient) and 15° for the amplitude and phase, respectively, and an S/N of 10 or better. The full FERG protocol was then started. FERG responses were acquired in sequence for the six values of modulation depth between 16.5% and 93.8%, presented in increasing order. Subjects were given a rest period of 30 to 45 seconds between stimulus presentations. The protocol ended with a blank noise recording. The total protocol lasted an average of 35 minutes. 

Data Analysis

For each subject or patient, the FERG-versus-modulation depth functions were evaluated off-line. After logarithmic transformation, FERG amplitudes were plotted as a function of the log stimulus modulation depth, and a linear regression (least-squares fitting) was fitted to the data points. Only amplitude values equal to or exceeding an S/N of 3 (see also later discussion) were included in the analysis. In all normal subjects and EL-ARM patients, all six data points of the FERG function satisfied the S/N criterion and were thus available. In all seven AL-ARM patients, the FERGs were not measurable (i.e., had an S/N <3) at the lowest modulation depth (16.5%), and in four of these seven patients responses were also not measurable at the second lowest modulation depth (33.1%). The number of FERG responses satisfying the S/N criterion are reported for each subject or patient in Table 1 . In all cases, the resultant functions were well represented by a straight-line relation (power function) with r ² ≥ 0.95. For each function, the slope was calculated and threshold estimated from the value of log modulation yielding a criterion log response amplitude corresponding to an S/N of 3. The criterion log amplitude was determined by taking three times the value of noise amplitude, obtained from the blank recordings (described earlier), and converting it to log units. Assuming that the noise is additive and the measured signal is actually signal plus noise, an S/N of 3 corresponds to a true S/N of 2 (for a similar approach in determining an S/N criterion, see also Reference 20). On average, the criterion amplitude was −0.85 ± 0.01 log μV (SE). The FERG amplitudes as well as the parameters obtained from each FERG modulation protocol—that is, the slope of the function and the threshold—were compared across control subjects and the two groups of ARM patients by analysis of variance (ANOVA), with post hoc tests (Tukey honestly significant difference [HSD]) for multiple comparisons. Response phase was recorded and plotted as a function of log stimulus modulation depth. Phases from normal subjects and EL-ARM patients were statistically compared by a two-way ANOVA with group (control subjects versus EL-ARM patients) as the between-subjects factor and modulation depth as the within-subjects factor. Phase data from AL-ARM patients were not included in the ANOVA because of missing data points. In all the analyses, P < 0.05 was considered statistically significant. 

Results

In Figure 1A , averaged FERG responses recorded from one control subject to the 41-Hz flicker stimulus presented at different modulation depths between 16.5% and 93.8%, are shown. Sweep duration corresponds to one stimulus cycle. Each response is the final average of eight blocks of 200 events each. Figure 1B reports the log amplitude and phase values of the FERG fundamental harmonic plotted as a function of log modulation depth. Error bars indicate the SE of amplitude and phase estimates at the fundamental component. With increasing log stimulus modulation depth, the increase in log amplitude was well described by a straight-line relation with a slope > 1, indicating nonlinear response gain. In all normal subjects, response slope was reliably above unity (see later description). FERG threshold was estimated from the value of log modulation depth yielding the criterion response. In the subject whose results are shown in the figure, FERG threshold was 1.19 (indicated by an arrow in Fig. 1B , top panel; 15.5% in linear units). Response phase as a function of log stimulus modulation depth showed relatively little change, with a phase advance (by approximately 20°) at the highest value of modulation depth. Both amplitude and phase displayed a small increase in variability at the lowest modulation depth. 

FERG amplitude and phase results, as well as a fluorescein angiogram obtained from a representative AL-ARM patient (AL-ARM patient 7 in Table 1 ), are shown in Figures 2A and 2B . In the plots of Figure 2A , the normal mean values (±SE) for amplitude and phase data are also reported for comparison. Mean amplitudes of control subjects as well as those of the patient have been fitted by linear regressions. The patient’s function was vertically shifted along the y-axis in relation to the normal mean, indicating a loss of response amplitude at all modulation depths. Patient’s FERG amplitude was not quantifiable (i.e., S/N < 3) at the lowest modulation depth. The slope of the patient’s function was reduced in relation to the normal mean values (logμ V/log % modulation depth: 1.15 versus 1.72 ± 0.08, normal mean ± SE;). Log FERG threshold, determined by linear extrapolation to criterion amplitude, was also increased in the patient compared with normal mean values (1.46 versus 1.18 ± 0.03, indicated by arrows in Fig. 2A , top). FERG phases of the patient were significantly delayed at the three lowest modulation depths (<1.9), compared with the normal mean values. 

In Figure 3A , group averaged (±SE) log amplitudes of the FERG fundamental harmonic are plotted as a function of log stimulus modulation depth for control subjects, AL-ARM patients, and EL-ARM patients. Compared with control subjects, EL-ARM patients showed significant (P < 0.05) losses in mean amplitudes at modulation depths higher than 1.6 (40%), whereas at lower values, mean amplitudes of both groups tended to overlap. The slope of the average amplitude versus modulation depth function was significantly (Tukey HSD, P < 0.01) shallower in EL-ARM patients (1.15 ± 0.05) compared with that of control subjects (1.72 ± 0.08, see also Table 2 ). AL-ARM patients had significantly (Tukey HSD, P < 0.01) smaller response amplitudes than those of normal subjects at all modulation depths and had no quantifiable responses (i.e., all S/N < 3) at values lower than 1.5 (32%). At the lowest recordable modulation depth (i.e., 1.52, 33.1%), AL-ARM patients had mean FERG amplitudes smaller (P < 0.05) than those of EL-ARM patients. The slope of the average function for the AL-ARM patients was significantly (Tukey HSD, P < 0.01) shallower (1.3 ± 0.12) than that of control subjects, whereas it did not differ significantly from that of EL-ARM patients. Figure 3B shows a scattergram of response function slopes versus log thresholds recorded individually in control subjects, EL-ARM patients, and AL-ARM patients. In the same plot, the lower and upper 95% confidence limits for the normal slope and threshold values, respectively, are also reported for comparison. In most (8 of 12) EL-ARM patients, response slope was lower than normal, whereas threshold was within the normal range. Only one patient of this group displayed an increased threshold, with a borderline slope. In six of seven AL-ARM patients, FERG thresholds were higher than normal. In three of these patients slopes were normal, whereas in three others the slopes were reduced below normal. The remaining AL-ARM patient had normal results for both slope and threshold. A summary of the abnormal FERG parameters found in each patient is reported in Table 1 . 

Average (±SE) slopes and thresholds of the FERG-versus-modulation depth functions recorded from control subjects and both groups of ARM patients are reported in Table 2 . ANOVA showed that either mean slopes (F[2,27]: 16.2) or thresholds (F[2,27]: 25.4) changed significantly (P < 0.001) across the groups of the study population. Tukey HSD tests indicated that mean slopes were significantly (P < 0.01) reduced, compared with control values, in both EL- and AL-ARM patients. Mean FERG thresholds did not differ significantly between control subjects and EL-ARM patients, whereas in AL-ARM patients mean thresholds were significantly (P < 0.01) increased compared with either normal or EL-ARM mean values. 

Figure 4A shows the FERG phase values, recorded individually from EL- and AL-ARM patients, plotted as a function of the log stimulus modulation depth. In the same plots, the mean and the lower 95% confidence limits (open circles and continuous lines, respectively) for the phase values recorded in normal subjects are also shown for comparison. Each patient within each group is represented by a separate symbol and line. Individual numbers correspond to those in Table 1 . Three EL-ARM patients showed abnormal FERG phase at all stimulus modulation depths, whereas three others displayed a normal phase at the highest modulation depth and delayed responses only at values lower than 1.8. In the remaining six EL-ARM patients, FERG phase was normal at all modulation depths. All seven AL-ARM patients displayed significant phase delays at one or more modulation depths lower than 1.9, whereas in five of seven patients, FERG phase was normal at the highest modulation depth. In Figure 4B , the group averaged (±SE) FERG phase values, recorded from control subjects, EL-ARM patients, and AL-ARM patients, are plotted as a function of log stimulus modulation depth. In both patients groups, mean FERG phases were delayed compared with those of control subjects. In EL-ARM patients, mean phases were delayed by approximately 60° (4 msec), at modulation depths lower than 1.8, with relatively small delays at the highest modulation depth. In AL-ARM patients, mean phases were delayed by 60° to 80° (4–5.4 msec) at all modulation depths. A two-way ANOVA, performed on the phase results obtained from control subjects and EL-ARM patients (data for AL-ARM patients were not included in the analysis because of unquantifiable values at the lowest modulation depth, see also the Materials and Methods section) showed a significant effect of group (control subjects versus EL-ARM patients, F[1,22]:7.6, P < 0.01) and a significant interaction of group by modulation depth (F[5,19]: 3.42, P < 0.05), indicating that phase delays observed in EL-ARM patients were dependent on the stimulus modulation depth. 

Discussion

In the present study, macular CFS was evaluated in normal subjects and ARM patients by using the FERG, a direct assay of outer retinal function. The relation between the response fundamental component and the modulation depth of the flicker stimulus was analyzed for amplitude and phase characteristics. In normal subjects, the response amplitude increased with stimulus modulation depth with a power law relation, with an exponent that was reliably greater than unity. This finding is in agreement with previous data ²⁰ showing a nonlinear gain in the FERG amplitude-versus-modulation depth function at temporal frequencies of approximately 40 Hz. Based on the FERG gain changes observed under different adaptation conditions, Wu et al. ²⁰ and Wu and Burns ²² suggested that there is an active nonlinear gain mechanism, controlling both amplitude and phase of the retinal flicker responses. The exact site of the retinal gain control is not unequivocally established yet, but the available evidence supports an outer retinal locus of the gain. In monkeys, Baron and Boynton ²¹ found that both the foveal local ERG and the aspartate isolated–late receptor potential increased in amplitude, as a function of the modulation depth of a 40-Hz flicker stimulus, with a slope higher than unity, suggesting the presence of a receptoral nonlinearity acting in a positive feedback manner. Valeton and Van Norren ³⁴ reported significant gain changes at the level of the cone photoreceptors by measuring both late receptor potential and local electroretinogram. In humans, Hood and Birch ³⁵ found changes in the gain and time constant of the a-wave of the flash ERG with light adaptation, supporting a cone photoreceptoral locus of the gain. Taken together, these previous findings are consistent with a model that places at least part of the retinal gain control mechanism at the level of cone photoreceptors themselves. However, a role of the retinal microcircuitry proximal to photoreceptors (bipolar and horizontal cells) also must be taken into account, given the experimental evidence ¹⁶ that bipolar cells substantially contribute to FERG generation. 

Results in patients showed that the function relating the FERG to stimulus modulation depth was significantly altered, compared with that of control subjects, with a different pattern of abnormalities depending on the severity of macular lesions. Response gain losses and modulation depth-dependent phase delays, with normal thresholds, were associated with early lesions. Increased thresholds, in addition to gain and phase abnormalities, were found in more advanced lesions. In the past, the FERG-versus-modulation depth function has been used to test retinal flicker threshold, gain, or both, in physiologic experiments ^{13 15 17 20 22} and in patients with retinitis pigmentosa or hereditary macular degeneration. ^{14 19 26} However, changes in these FERG parameters have not been investigated in ARM. The present results suggest a pathophysiological sequence in which, in association with early Bruch’s membrane and RPE changes (i.e., soft drusen), there are already some signs of retinal dysfunction involving response gain. Indeed, amplitude gain losses and modulation depth–dependent phase delays could be the expression of the same altered control mechanism. These abnormalities may result from early degenerative changes ³ of cone photoreceptors, whose number at this stage is presumably normal or near normal. Shortening of cone outer segments, by reducing quantum catch and therefore the effective retinal intensity of the stimulus, and altered photoreceptor membrane properties (i.e., time constants ^{26 36} ) by delaying reestablishment of equilibrium, may both affect retinal gain. As the disease spreads to larger areas, with a spatially dependent loss ² of photoreceptors, retinal dysfunction becomes more marked and manifests itself with an increase in threshold. It has been indeed suggested ¹⁴ that FERG sensitivity losses may result from photoreceptor dropout, either spatially dependent or independent, assuming that detection depends on the pooled contribution of underlying retinal elements. A simpler explanation of the present findings may be that different degrees of severity of cone system dysfunction, independent of the number of remaining photoreceptors, are reflected first by changes in response gain and phase and then by an increased threshold. Clearly, the proposed sequence of FERG abnormalities, whatever the underlying mechanism would be, should be validated by longitudinal studies with clinical evaluations and serial recordings in the same ARM patients. 

It may be of interest to compare the present data with previous electrophysiological and psychophysical findings documenting cone system dysfunction in early ARM. FERGs have been evaluated in early ARM eyes that, unlike the eyes tested in the present study, were the fellow good eyes of patients with unilateral neovascular macular degeneration. ^{37 38} Although none of the previous studies analyzed the FERG-versus-modulation depth function, and the patients tested may have had a more severe type of ARM, increased phase delays, compared with normal responses, were found in association with large drusen and choroidal filling defects in the macular area. ³⁸ Together with the present findings, these data suggest that the temporal characteristics of foveal cone photoreceptors are altered in association with early metabolic changes in the choriocapillaris and Bruch’s membrane. CFS was found to be altered in eyes affected by early ARM. ^{5 8 11 12} As in the FERG studies, the eyes tested belonged to patients with unilateral exudative macular degeneration and cannot be directly compared with the eyes in the present study. Nevertheless, significant losses in flicker sensitivity, in comparison with healthy aging eyes, were observed in eyes with typical ARM lesions and normal visual acuity. ⁵ Sensitivity losses of 0.2 to 0.4 log units involved mainly the midfrequency range (10–40 Hz) and, according to a model proposed by Mayer et al., ⁵ were attributed to a reduction in sensitivity of the high-frequency mechanism underlying psychophysical CFS. Although the temporal modulation sensitivity assessed by FERG may have different characteristics compared with the psychophysical CFS, ¹³ the amount of FERG sensitivity losses found in our AL-ARM patients appears to be consistent with the previously found psychophysical losses. 

In summary, the results of this study show that, in early ARM, FERG response gain and phase characteristics may be affected without appreciable changes in modulation depth thresholds. In more advanced stages, FERG thresholds tend to increase in relation to normal values, indicating loss of retinal sensitivity. Abnormalities of response gain and phase suggest an altered retinal gain control mechanism that may reflect abnormalities in cone photoreceptor function (quantum catching ability, temporal response properties), occurring early in the disease process. An increased threshold may result from more severe cone dysfunction, loss, or both, characteristic of an advanced stage. The present approach shows potential clinical value to directly document different stages of macular cone dysfunction in ARM. It may also provide a more complete set of parameters, compared with other FERG techniques based on maximum response recordings, to monitor macular function during treatments with potential therapeutic agents. 

 Submitted for publication June 15, 1999; revised September 15 and November 17, 1999; accepted December 1, 1999.

 Commercial relationships policy: N.

 

Table 1.

 

Demographic and Clinical Findings in Normal Subjects and Patients with ARM

	Age, Sex	Acuity	Fundus*	Drusen Size, †	RPE Abnormalities, ‡	Pupil, §	FERG Abnormal Parameter, ∥
Controls
1	54, F	20/20	Normal	—	—	8.5	(6)
2	65, F	20/20	Normal	—	—	9	(6)
3	64, M	20/20	Normal	—	—	9	(6)
4	64, F	20/20	Normal	—	—	9	(6)
5	64, F	20/20	Normal	—	—	8.5	(6)
6	65, F	20/20	Normal	—	—	9	(6)
7	84, M	20/20	Normal	—	—	9	(6)
8	65, M	20/20	Normal	—	—	8.5	(6)
9	64, M	20/20	Normal	—	—	9.5	(6)
10	70, F	20/20	Normal	—	—	8	(6)
11	64, F	20/20	Normal	—	—	9	(6)
EL-ARM
1	60, F	20/20	Soft drusen, middle subfield	9; ≥63, <125	—	9	Slope, phase, (6)
2	68, F	20/25	Soft drusen, middle subfield	6; ≥63, <125	—	9	Phase, (6)
3	71, F	20/20	Soft confluent drusen, middle subfield	7; ≥125, <175	—	8.5	(6)
4	84, M	20/25	Soft and hard drusen, central and middle subfield	10; ≥63, <125	—	8.5	Slope, phase, (6)
5	80, M	20/20	Soft drusen, middle subfield	10; ≥63, <125	—	9	Slope, (6)
6	65, F	20/30	Soft drusen, central and middle subfield	12; ≥125, <175	—	8.5	Threshold, slope (borderline), (6)
7	64, F	20/20	Soft drusen, middle subfield	8; ≥63, <125	—	9	Slope, phase, (6)
8	67, M	20/20	Soft confluent drusen, middle subfield	12; ≥125, <175	—	8.5	Phase, (6)
9	60, F	20/25	Soft confluent drusen, central and middle subfield	15; ≥125, <175	—	9	Slope, phase, (6)
10	61, F	20/20	Soft drusen, middle subfield	6; ≥63, <125	—	9	Slope, (6)
11	66, F	20/30	Soft drusen, central and middle subfield	8; ≥63, <125	—	9	Slope, (6)
12	54, F	20/25	Soft drusen, central and middle subfield	12; ≥63; <125	—	9	Slope, (6)
AL-ARM
1	63, M	20/30	Soft drusen; hyperpigm., middle subfield	25; ≥63, <125	30	9	Threshold, phase, (4)
2	81, F	20/30	Soft confluent drusen, central and middle subfield	28; ≥175, <250	—	8.5	Threshold, phase, (4)
3	71, F	20/25	Hyperpigm. and hypopigm., middle subfield	—	40	8.5	Threshold, slope, phase, (4)
4	64, F	20/25	Soft confluent drusen, hypopigm., middle subfield	22; ≥125, <175	30	9	Phase, (5)
5	66, M	20/20	Hyperpigm., hypopigm., middle subfield	—	40	9	Threshold, phase, (5)
6	64, M	20/20	Soft confluent drusen, hyperpigm., middle subfield	24; ≥125, <175	45	9	Threshold, slope, phase, (4)
7	66, F	20/25	Hypopigm. and hyperpigm., middle subfield	—	40	9	Threshold, slope, phase, (5)

*  Macular appearance including drusen type, confluence, and location; RPE abnormalities type and main location.

†  Drusen number and size (in micrometers) according to the international classification and grading system.

‡  Percentage of the main middle subfield area affected with RPE abnormalities.

§  Pupil diameter (in millimeters), pharmacologically dilated, for the FERG recordings.

∥  FERG function parameter (threshold, slope, phase) that was outside normal 95% confidence limits in an individual patient. Includes the number of FERG responses that were above noise level (i.e., S/N ratio≥ 3) at the different modulation depths of the recording protocol: (6) = S/N ratio ≥3 at all modulation depths; (5) = S/N ratio<3 at the lowest modulation depth; (4) = S/N ratio <3 at the two lowest modulation depths.

Figure 1.

 

(A) Representative averaged FERGs recorded from a control subject in response to 41-Hz flicker stimulus presented at different modulation depths between 16.5% and 93.8%. (B) Log amplitude and phase values of the FERG fundamental harmonic plotted as a function of log modulation depth. Error bars indicate the SE of amplitude and phase estimates at the fundamental component. (B, arrow, top) Extrapolated threshold for this subject.

Figure 2.

 

(A) Log FERG amplitudes and phases (▴) recorded at different values of log stimulus modulation depth in a representative patient with an AL-ARM lesion (AL-ARM patient 7 in Table 1 ). Normal mean values (±SE) are also shown for comparison. Amplitude data have been fitted by linear regressions. Arrows on the x-axis indicate the log FERG thresholds, determined for the average control subjects’ function (leftmost arrow) and the patient’s function. (B) Fluorescein angiogram of the same patient.

Figure 3.

 

(A) Group-averaged (±SE) log amplitudes of the FERG fundamental harmonic plotted as a function of log stimulus modulation depth for control subjects, EL-ARM patients, and AL-ARM patients. (B) Scattergram of response function slopes versus log thresholds recorded individually from control subjects, EL-ARM patients, and AL-ARM patients. Dashed lines in the plot indicate the lower and upper 95% confidence limits for the normal slope and threshold values, respectively.

Table 2.

 

FERG Function Slopes and Thresholds Recorded in the Different Groups of the Study Population

Groups	Slope	Threshold
Normal subjects (n = 11)	1.72 ± 0.08	1.18 ± 0.03
ARM patients, early lesion (n = 12)	1.15 ± 0.05*	1.11 ± 0.04
ARM patients, advanced lesion (n = 7)	1.3 ± 0.12*	1.53 ± 0.06* , †

 Data are mean ± SE. Slope is log μV/log percentage modulation depth and threshold is log percentage modulation depth.

*  Significantly different (P < 0.01) from the normal mean value.

†  Significantly different (P < 0.01) from the early lesion mean value (Tukey HSD test).

Figure 4.

 

(A) FERG phases recorded individually from patient with EL-ARM (top) and those with AL-ARM (bottom) at different values of log stimulus modulation depth. Open circles and solid lines in the plots indicate the mean and the lower 95% confidence limits, respectively, established for the normal values. (B) Group-averaged (± SE) FERG phase values, recorded from control subjects, EL-ARM patients, and AL-ARM patients plotted as a function of log stimulus modulation depth.