Regional Cone Dysfunction in Retinitis Pigmentosa Evaluated by Flicker ERGs: Relationship with Perimetric Sensitivity Losses

Abstract

purpose. To evaluate regional cone dysfunction in retinitis pigmentosa (RP) by recording focal electroretinograms (FERGs) from the central and paracentral retinal regions and to correlate the FERG with perimetric sensitivity losses.

methods. Twenty-three typical patients with RP (age, 18–65 years; visual acuity, 20/100 to 20/20; kinetic visual field by size II/4e, 20–40°) and eight age-matched control subjects were evaluated. FERGs were recorded in response to either a central (eccentricity, 0–2.25°) or a paracentral annular (2.25–9°) field, presented on a light-adapting background. Fields’ luminances (mean: 80 cd/m²) were sinusoidally modulated at different temporal frequencies (TFs; 10.3, 14, 21, 32, 41, and 52 Hz). Amplitude and phase of the responses’ fundamental harmonic (1F) were measured. Perimetric sensitivity was measured by a visual field perimeter. For each patient, mean sensitivity losses were calculated for both the central (0–2.25°) and paracentral (2.25–9°) regions.

results. On average, central and paracentral FERGs of patients with RP were reduced in amplitude (P ≤ 0.05) compared with control values. Amplitude losses tended to be smaller in the central than the paracentral region and were limited to low-medium TFs (10.3–14 Hz). Paracentral losses were rather invariant with TF. Paracentrally, but not centrally, the FERG phase in patients was delayed on average (P < 0.01), compared with control values. The central FERG phase was delayed only in patients with visual acuities less than 20/40. In individual patients, paracentral 41-Hz amplitude losses were positively correlated with corresponding perimetric losses (r = 0.7, P < 0.005). Both central and paracentral 41-Hz amplitudes displayed high specificity (87.5% and 100%, respectively) with relatively low sensitivity (46.6% and 63.6%, respectively) in predicting perimetric results in corresponding retinal regions.

conclusions. In RP, central and paracentral FERGs are differently altered as a function of TF, indicating regional differences in the stage and/or pathophysiology of retinal cone dysfunction. FERG abnormalities may predict, to some extent, perimetric results at corresponding retinal regions. The data support the use of the present FERG method to evaluate regional cone dysfunction in different stages of RP.

Retinal cone-mediated function can be altered from the early stages of retinitis pigmentosa (RP). Involvement of both cone photoreceptors and postreceptoral retina has been well characterized by anatomic studies in human donor RP retinas ^{1 2} and by clinical studies in patients. ^{3 4} Both macular and peripheral cone function in patients with RP may show abnormalities in temporal responsiveness, ⁵ decreased flicker sensitivity, ^{5 6 7} and changes in response timing. ⁸ Typically, however, central function is more preserved than peripheral function, when evaluated by psychophysical or electroretinographic techniques. ^{9 10 11 12} It has been suggested ⁹ that such a difference is related to the different density of photoreceptors, which degenerate at a constant rate, between central and peripheral retinal regions. 

Regional variations of cone function can be evaluated by focal electroretinogram (FERG), ^{7 13} a response generated by stimuli presented to localized retinal areas on a steady adapting background, used to minimize stray-light modulation. So far, relatively few studies have evaluated FERG responses at discrete retinal areas or have correlated the electrical activity with perimetric sensitivity at corresponding locations. Indeed, most of the studies on the correlation between ERG and psychophysics in RP evaluated results obtained by using full-field ERG. ^{14 15 16 17 18} This technique, however, because of the nature of the stimulus, is not suited to analyze the relationship between retinal dysfunction and psychophysical loss at discrete retinal areas. More recently, Seeliger and Kretschmann ¹⁰ and Hood et al., ¹¹ by recording the multifocal ERG, found greater amplitude reductions of the paracentral than the central responses, and delayed implicit times were found only in the paracentral responses. Hood et al. also reported that abnormal timing, rather than amplitude reduction of local FERGs, can be a good predictor of corresponding perimetric losses. 

An alternative approach to evaluating regional variations of cone function is represented by FERG recordings of responses to sinusoidally flickering stimuli. ¹³ Although in the multifocal technique each local response represents a single response function derived from cross-correlation analysis, ¹⁹ in the flicker technique the ERG signal at a given temporal frequency (TF) can be directly analyzed in its amplitude and phase. In the past, investigators have evaluated the physiological properties of flicker ERGs recorded from the foveal and extrafoveal regions, ^{20 21 22 23} providing evidence that foveal responses display different timing and sensitivity characteristics when compared with responses from more eccentric retinal regions. Knowledge about local changes in the flicker ERG may be relevant to the understanding of abnormalities in the full-field flicker ERG, a signal commonly used to monitor cone function in RP (see, for example, Ref. ²⁴ ). The flicker ERG approach could also help to evaluate the TF response function and to detect selective deficits in amplitude or phase. It has been used to characterize macular dysfunction in patients with RP and in patients with hereditary macular degeneration. ¹³ Both psychophysical ^{8 25} and electrophysiological ^{13 26 27} studies have demonstrated that patients with RP may show specific abnormalities of the photopic temporal processing, revealed by a change in the temporal flicker sensitivity function or in the electroretinographic temporal response function. 

In the present study, the FERG technique first used by Seiple et al. ¹³ was applied to evaluate central and paracentral retinal functions in patients with typical RP. The intent was to extend and refine the results of Seiple et al. by evaluating, in patients with good central fixation and variable visual acuity, the FERGs as a function of stimulus TF at both central and paracentral retinal locations. Another purpose of the present work was to evaluate the relationship between regional variations in FERG responses and cone-mediated perimetric sensitivity losses at corresponding retinal regions. The results indicated that, in patients with RP, central and paracentral responses may be differently altered as a function of TF, and losses in both responses show a relationship with corresponding perimetric losses. 

Materials and Methods

Subjects

Twenty-three patients (10 males and 13 females; mean age, 37 years; range, 15–64) with a diagnosis of typical RP were included in the study. Each patient underwent a complete general and ophthalmic examination. A detailed family and medical history was also obtained for each. Thirteen of the patients were categorized as having the simplex (sporadic) form of RP, three had a family history suggesting a dominant mode of transmission, and seven were categorized as having autosomal recessive type. Each patient had severely reduced or nondetectable standard full-field ERGs (both scotopic and photopic, recorded according to the International Society for Clinical Electrophysiology of Vision [ISCEV] standard protocol ²⁸ ), a central kinetic visual field of at least 20° (measured by Goldmann perimetry with a II/4e white target), and a Snellen acuity between 20/20 and 20/100. On fluorescein angiography, none of the patients had evidence of cystoid macular edema. All patients included in the study had stable central fixation (as evaluated by a Visuskope; Heine, Germany), clear optical media, and no concomitant ophthalmic diseases. Clinical details of individual patients with RP are reported in Table 1 . Eight normal subjects (four males and four females; mean, 35; range, 12–62) provided normative FERG data. None of the subjects had a history of ophthalmic or neurologic disease. All subjects had normal general and ophthalmic examinations. Informed consent to participate in the study was obtained from all subjects, and the research complied with the tenets of the Declaration of Helsinki. 

Electrophysiological Methods

The stimuli consisted of flickering uniform fields generated by an array of 8 red LEDs (max λ, 660 nm; half-height bandwidth, 25 nm; mean luminance, 80 cd/m²) sinusoidally driven by a custom-made digital frequency generator, ²⁹ and presented on the rear of a Ganzfeld bowl, illuminated at the same mean luminance as the stimulus. A diffusing filter placed in front of the LED array made it appear as a circle of uniform red light. A steady direct-current (DC) signal maintained the constancy of the stimulus mean luminance. FERGs were recorded in response to the sinusoidal luminance modulation (92% modulation depth) of two different homogeneous fields presented in the macular region: a circular field 4.5° in diameter and an annular field with an outer and inner diameter of 18° and 4.5°, respectively. A small central fixation mark allowed both fields to be centered on the fovea. Patients and normal subjects always fixated monocularly at the fixation mark, from a distance of 30 cm. Under these conditions, the circular and annular fields stimulated the central (0–2.25° eccentricity) and the paracentral (2.25–9° eccentricity) retinal regions, respectively. Stimulus TF was varied between 10.3 and 52 Hz. In both control subjects and patients, pupils were pharmacologically (1% tropicamide and 2.5% phenylephrine hydrochloride) dilated to at least 8 mm. 

FERGs were recorded by an Ag-AgCl electrode taped on the skin over the lower eyelid. A similar electrode, placed over the eyelid of the contralateral patched eye, was used as reference (interocular ERG, ³⁰ ). Because the recording protocol was extensive, the use of cutaneous electrodes with interocular recording represented a good compromise between signal-to-noise ratio and signal stability. FERG signals were amplified (100,000-fold), band-pass filtered between 1 and 100 Hz (6 dB/octave), and averaged (12-bit resolution, 2-kHz sampling rate, 200–600 repetitions in two to six blocks). The averaging time (i.e., the sweep duration) was varied according to the stimulus period. Single sweeps exceeding a threshold voltage (5 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 (1F) and measure its peak-to-peak amplitude (in μV) and phase (in degrees). Averaging and Fourier analysis were also performed on signals sampled asynchronously at 1.1 times the TF of the stimulus, to give an estimate of the background noise at the stimulus frequency. Under the present experimental conditions, the FERGs recorded individually from both control subjects and patients with RP were above the noise level (noise amplitude ≤0.08 μV in all cases) and sufficiently reliable (the variation coefficient in amplitude was typically 20%; the phase standard deviation was ±20°). 

Automated Perimetry

Static perimetric thresholds in patients with RP were obtained with a visual field perimeter (10-2 program, Statpac software; Humphrey Instruments, San Leandro, CA), which tests 49 retinal locations of the macular region. Points exceeding 9° in eccentricity were excluded from the analysis, so that retinal areas tested with FERG and perimetry corresponded exactly. Mean sensitivity losses, in relation to age-matched data provided by the perimeter’s software, were calculated separately from the total-deviation plot, for the central and paracentral areas. In addition, sensitivity losses were separately plotted as a function of retinal eccentricity for different field meridians (0–180°, 90–270°, 45–225°, and 135–315°). Patients’ perimetric data were then compared with the normative data collected in our laboratory from 10 normal subjects, age-matched with the patients. Eight of these 10 normal subjects had also participated in the FERG study. 

Statistical Analysis

FERG and perimetric results in one eye of each patient, typically the eye with the best visual acuity, were included in the analysis. When the two eyes had the same visual acuity, the test eye was randomly selected. The results in one randomly selected eye of each normal subject were also included. FERG 1F amplitudes of central and paracentral responses from normal subjects and patients with RP were statistically compared by two-way analysis of variance (ANOVA), with group (normal subjects versus patients) as the between-subjects factor and TF as the within-subject factor. Because previous studies ¹³ have shown that the magnitude of FERG amplitude losses at high TFs in patients with RP may depend on visual acuity, subgroup analyses of data were performed in patients with acuity better or worse than 20/40. FERG amplitudes were expressed as response density (amplitude/stimulus area, in nanovolts per degree squared) to normalize them in relation to the different cone photoreceptor densities in central and paracentral regions. Response amplitudes also underwent logarithmic transformation for better approximation of a normal distribution. FERG phase values were averaged, and corresponding variances (and circular standard deviations) were estimated, using a method that takes into account the circular distribution of phase space, ³¹ after conversion of amplitude and phase data into cosine and sine values. The stimulus-response phase differences associated with TFs from 10.3 to 52 Hz were calculated ¹³ using the relationship: phase = (response lag/stimulus period) × 360°. Because the Fourier analysis gives only the response phases in a 360° range, and the actual phase values can be, in theory, integer multiples of 360°, several assumptions were made to determine the exact response phases as a function of TF in both control subjects and patients. First, it was assumed that the phase of the various FERG responses is mostly determined by a time delay, which is comparable to the implicit time of the standard cone flicker ERG. ³² Second, it was assumed that, whereas in normal subjects FERG timing would comprise between 25 and 40 ms, ^{32 33} in patients with RP, response timing may be greatly increased compared with normal values, as shown for the photopic flicker ERG ³² or for the fundamental component of the flicker ERG in the range 14 to 52 Hz (up to 33-ms increase from normal mean values ²⁷ ). Therefore, timing of our FERG measures in normal subjects and patients was expected to be between 25 and 73 ms. Third, it was assumed that response phase lags linearly with TF in the range 10 to 52 Hz ^{13 27} and that mean slopes of phase-lag as a function of TF may provide an estimate of the overall response latency. ¹³ Slopes of phase-lag versus TF were therefore computed for both central and paracentral FERGs, and compared between groups by t-tests. Correlation between individual FERG amplitude losses at different TFs for the central and paracentral areas and perimetric sensitivity losses at corresponding retinal eccentricities was evaluated by Pearson’s correlation and linear regression analysis. For the main analyses, P < 0.05 was considered statistically significant. For multiple comparisons, a more conservative P < 0.01 was adopted. 

Results

Regional Variations of FERG Responses

In Figure 1A , the retinal location of the stimulus for both central and paracentral recordings is shown schematically. FERG responses obtained from the right eye of a control subject at the temporal frequency of 41 Hz for both central and paracentral stimuli are depicted in Figure 1B . Each record in Figure 1B represents the final average of eight blocks of 200 events each. In the same figure , the Fourier-isolated 1F components of both responses have been superimposed on the raw records. One full response cycle in the 24-ms epoch is shown for either central or paracentral response. To the right of each record, scatterplots of cosine and sine values of the individual blocks of FERG 1F components are also shown. Each data point represents a vector with a length that reflects the response amplitude and orientation the phase angle. Phase delay is in a clockwise direction. It can be noted that the phase of the Fourier-isolated 1F of the central response is slightly but consistently delayed, compared with that of the paracentral response. In Figure 1C , the 41-Hz raw waveforms depicted in Figure 1A have been normalized (see the Materials and Methods section) to the stimulus area, to provide the corresponding FERG response densities. As expected for a focal electroretinographic response, ²¹ the response density was greater to the central than the paracentral stimulus. 

Mean FERG amplitudes (±SEM), expressed as response densities, recorded at different TFs from normal subjects and patients in response to central and paracentral stimuli are plotted in Figure 2 . Data for patients with RP are presented separately for the subgroups with visual acuity better (n = 16) or worse (n = 7) than 20/40. It can be seen that, at all the tested TFs, the response density in control subjects was greater to the central than the paracentral stimuli. It can also be noted that, in agreement with previous data on the TF amplitude spectrum of the 1F component of the FERG, ^{21 22 27 34 35} both temporal functions peaked at 41 Hz, both in control subjects and in patients with RP. Mean amplitudes in patients were smaller than those in control subjects at both retinal locations. In general, amplitude losses in patients tended to be greater in response to the paracentral than the central stimulus. For the central but not the paracentral stimulus, amplitude losses appear to be more pronounced at low (10.3–14 Hz) than at high TFs. A two-way ANOVA showed that, for the central stimulus, there was a significant effect of group (F_(5,33) = 1.27; P = 0.029) and TF (F_(5,33) = 8.08; P < 0.001). However, the interaction of group by TF did not reach the significance level (F_(5,13) = 0.58; NS). Similarly, for the paracentral stimulus, there was a significant effect of group (F_(1,27) = 25.05; P < 0.001) and TF (F_(5,13) = 8.88; P < 0.001), but no significant interaction of group by TF (F_(5,13) = 1.76; NS). Adjusted post hoc t-tests showed that, with the central stimulus, mean amplitude reductions in patients, compared with those in control subjects, were significant only at 10.3 and 14 Hz (P < 0.01). With the paracentral stimuli, the mean 1F amplitudes of patients were significantly reduced, compared with those in control subjects, at all (P < 0.01) but the lowest (10.3 Hz) TF. No significant differences were found for both the central or paracentral response amplitudes between patients with acuities better or worse than 20/40. 

Average (± circular standard deviations, see the Materials and Methods section) FERG phase values, recorded as a function of TF for both central and paracentral stimuli in normal subjects and the whole group of patients with RP, are shown in Figure 3A and 3B (top). It can be noted that for the central stimulus, response versus TF phase functions of both groups were similar. In both normal subjects and patients, the average phase lagged linearly with temporal frequency (least-square fitting, r = −0.98 and −0.95, respectively). The average slope was 8.8° ± 2.8° (SD)/Hz for normal subjects (close to that previously reported by Seiple et al., ¹³ with a 9° stimulus) and 6° ± 2°/Hz for patients (not significantly different from the normal value by t-test). With the paracentral stimulus, average phase functions appeared to be greatly different, when comparing normal with patient data, with the latter showing an increasing delay with TF. In both normal subjects and patients, the average phase lagged linearly with TF (r = −0.98 and −0.97, respectively). The average slope was 6.6 ± 2.7 deg/Hz for normal subjects and 15.2 ± 2.8 deg/Hz for patients. The two values differed significantly by t-test (t = 7.5, df 30; P < 0.01). In Figure 3 (bottom panels) average phase data of both central and paracentral stimuli are presented separately for patients with acuities better or worse than 20/40. It can be noted that for the central stimulus, phase delays at high TFs were greater in patients with acuities worse than 20/40 than in those with better acuities. The average phase slope was 14.4 ± 2 deg/Hz in the former and 6.6 ± 2 deg/Hz in the latter. The two values differed significantly by t-test (t = 8.6; df 22; P < 0.01). For the paracentral stimulus, average phase values as a function of TF did not differ significantly between patients’ subgroups (mean slopes of 15.5 ± 1 deg/Hz and 15.2 ± 1.2 deg/Hz, respectively). 

For both central and paracentral stimuli, the RP patients’ phase distributions at the different TFs showed an increased interindividual variance, compared with the normal distributions. These differences could be related to the lower signal-to-noise ratios of the patients’ responses. Indeed, an increased noise influence on the responses may be a source of phase variability (see Ref. ³⁶ for a recent review), resulting in an increased scatter of cosine and sine response values. For better evaluation of this effect, the responses at all TFs were compared with the corresponding noise levels in each patient. There was no tendency toward an increased phase scatter in the individual responses with low compared with those with high signal-to-noise ratios, suggesting that this factor probably did not account for the increased phase variance. In addition, several patients’ responses with phases greatly different from the average normal values showed normal or near normal signal-to-noise ratios. This finding is illustrated in Figure 4 , where cosine and sine values of 10.3- and 41-Hz 1F components, recorded from the central retinal region, are plotted for normal subjects and patients, together with the corresponding individual noise values. It can be seen that, at both TFs, there were patients’ responses with vector lengths that were well above the noise distribution, as for normal responses, yet with large differences in orientation (i.e., most presumably, phase delays) compared with normal values. Hence, we concluded that the increased phase variability found in our patients was not the result of a generally lower signal-to-noise ratio, but rather were an effect of disease per se, which produced in some patients an abnormally delayed response phase, resulting in an increased scatter of values for the whole population. 

Relationship between FERG and Perimetric Losses in Patients with RP

Table 1 reports the perimetric sensitivity losses recorded in both the central and paracentral retinal regions in each patient. Mean sensitivity losses were significantly greater in the paracentral than the central region (t = 9.9, df 32, P < 0.01). Central sensitivity was relatively preserved (<5 dB loss) in several patients. ANOVA did not reveal any statistically significant variation of perimetric loss across field meridians. An independent t-test comparing the results from control subjects and patients was statistically significant for both the central (t = 6.5; P < 0.001) and the paracentral (t = 8.7; P < 0.001) stimuli. Pearson’s correlation showed that perimetric sensitivities at central and paracentral locations (see the Materials and Methods section) were positively correlated with corresponding central and paracentral FERG amplitudes recorded at 41 Hz. No significant correlations were found between perimetric sensitivities and ERG amplitudes recorded at the other frequencies, or with FERG phase data. For instance, 41-Hz phase delays were found in some patients with relatively preserved perimetric sensitivities (<5 dB loss), whereas in other patients with substantial perimetric loss, response phase was normal. In Figures 5A and 5B , perimetric sensitivity losses, recorded individually in patients with RP for the central (Fig. 5A) and paracentral (Fig. 5B) stimuli, are plotted as a function of the corresponding 41-Hz FERG response density losses. For both parameters, log data are plotted. The lower 95% normal confidence limits for response density and perimetric sensitivity are also shown. The correlation between perimetric and ERG measurements was statistically significant with both the central (r = 0.4; P < 0.05) and the paracentral (r = 0.7; P < 0.005) stimulus. However, the association was clearly stronger with the latter than with the former. Regarding the central responses, the influence of one or more outliers in determining the correlation cannot be excluded. From the analysis of the scatterplot in Figure 5A , it can be noted that only 1 of 23 patients showed a reduced central response amplitude with normal perimetric values, whereas 8 of 23 patients had normal response density but altered perimetric thresholds. For the paracentral stimulus (Fig. 5B) , none of the patients with RP showed a reduced ERG response amplitude associated with normal perimetric sensitivity, whereas 8 of 23 patients had normal ERG amplitude with reduced perimetric sensitivity. The diagnostic sensitivities, specificities, and accuracies of 41-Hz central and paracentral FERG amplitudes in predicting, in individual patients, perimetric sensitivity results (normal or abnormal) at corresponding retinal locations are reported in Table 2 . It can be seen that both central and paracentral amplitude losses displayed high specificity, with relatively low sensitivity, in predicting corresponding perimetric results. In other words, patients with normal central or paracentral FERG amplitudes were likely to have normal field findings, whereas the ability to predict an abnormal field, given an abnormal FERG amplitude, was much lower. 

Discussion

In this study, regional variations of cone function were evaluated in patients with RP by recording central (4.5°) and paracentral (2.25–9° retinal eccentricity) FERGs as a function of TF in the range 10.3 to 52 Hz. FERG amplitudes of patients were compared with perimetric sensitivities measured at corresponding retinal locations. The results showed amplitude losses and phase delays in both central and paracentral ERGs. As expected, paracentral amplitude losses were greater than corresponding central losses, in agreement with the clinical picture and histologic studies ³⁷ showing a greater preservation of retinal cell populations in the macula than in the extramacular regions. For the central ERG, mean amplitude losses in patients were limited to low-medium (10.3–14 Hz) frequencies. For the paracentral ERG, mean amplitude losses were substantial at all but the lowest (10.3 Hz) TF. At high TFs, phase delays were greater with the paracentral than the central stimuli. Significant central delays were observed only in patients with acuity worse than 20/40. By contrast, paracentral stimuli revealed phase delays independent of patients’ acuity or field sensitivity. In individual patients, response delays tended to be associated with either normal or reduced amplitudes and therefore did not result from an increased influence of the noise on the response. Both central and paracentral amplitudes were significantly correlated with corresponding perimetric sensitivities. 

The present results are directly relevant to other psychophysical and FERG data previously reported in the literature on macular flicker responses in patients with RP. Psychophysical studies have found in the central (foveal) region both low- and high-frequency sensitivity losses, ^{38 39} only high TF losses, ⁴⁰ or an overall sensitivity loss without change in the corner frequency and critical fusion frequency. ²⁵ FERG results from Seiple et al. ¹³ showed both low and high TF losses, with the magnitude of the latter depending on the patient’s visual acuity. Seiple et al. ¹³ also reported no significant differences in the phase-lag as a function of TF between normal subjects and patients with RP. The current data are partially at odds with these previous reports. However, a combination of differences in methodology and sample selection may prevent a direct comparison of the data. Psychophysical flicker sensitivity, exploring threshold visual responses, may not be strictly correlated with FERG suprathreshold signals. In the study by Seiple et al.. ¹³ a larger central retinal area (9°) compared with that tested in the present study, was evaluated by FERG. Indeed, the current results would probably have shown both low and high TF losses had the central and paracentral areas been sampled simultaneously with the same high-intensity stimulus (i.e., the paracentral amplitude losses at high TFs may have influenced the average shape of the frequency response function). It should be considered that the present study differed from that of Seiple et al. ¹³ for other important methodological aspects. For instance, the average stimulus luminance used by Seiple et al. was 50 cd/m², lower than that used in this study (80 cd/m²). In addition, they evaluated the peak-to-peak amplitude and phase of the whole FERG signal, whereas in the present study the fundamental harmonic component of the response, isolated by Fourier analysis, was measured. It is well established, ^{27 41 42} and also apparent from Figure 2 in Seiple et al., ¹³ that a significant distortion of the ERG waveform produced mainly by a second harmonic component is present at frequencies between 10 and 30 Hz. The second harmonic data were not evaluated in the present study, because a full analysis of the second-harmonic–response function would have required lengthy recording sessions to collect responses at TFs lower than 10 Hz, at which the second harmonic has its maximum amplitude in normal subjects. ²⁷ Taking into account the whole FERG response instead of the 1F component may result in a significant difference in the shape of response amplitude and/or phase versus TF functions. Finally, it is possible that the distribution of individual patients’ visual acuities may have differed between the present sample and that tested by Seiple et al., ¹³ thus precluding a direct comparison of the results. 

The present findings also differ, at least in part, from those obtained by using multifocal ERG, ^{11 12} where a delay in implicit time was found only for the paracentral responses and with previous studies reporting normal foveal ERG implicit times in patients with RP with delayed full-field ERGs. ^{21 35} Significant changes in the phase-versus-TF function of central ERG responses, indicating response delay, were indeed observed in our patients, although only in the group with visual acuity worse than 20/40. It is possible that delays in the flicker ERGs recorded from the central region occur only when the cone system undergoes more advanced stages of degeneration. The FERG 1F component is known to reflect mainly the activity of cone bipolar cells ^{43 44} with a contribution of cone photoreceptors. ^{13 42} Altered response properties (i.e., temporal integration) of cone photoreceptors or bipolar cells or selective deficits in these generators’ subpopulations sensitive to specific TFs may contribute to the observed abnormalities. Assuming that the cone-system degeneration process in RP is similar in both central and eccentric locations, ⁹ a slowing of the ERG signal could be expected, regardless of retinal eccentricity. Therefore, the differences found between central and paracentral delays could simply reflect a different stage of retinal degeneration. Alternatively, the observed difference may reflect a regional specificity in the pathophysiology of cone dysfunction. The different TF dependence of amplitude losses observed by comparing central with paracentral FERGs in our patients lend support to the last hypothesis. From a clinical perspective, a significant central response delay associated with amplitude losses at low TFs, recorded in an individual patient with RP, could be considered as a sign of advanced central cone-system dysfunction in the course of the disease process. 

A finding previously reported by Hood et al., ¹¹ who used multifocal ERG, was that amplitude loss was not a good predictor of visual sensitivity at the Humphrey visual field. By contrast, they reported a strong correlation between implicit time of the local responses and corresponding perimetric sensitivities. In the present study a significant correlation between FERG amplitude and visual field sensitivity losses was found for both central and paracentral stimuli. The association was relatively stronger with the latter stimuli. Clinically, the FERG was highly specific in predicting normal perimetric results, both in the central and paracentral locations. Diagnostic sensitivity, however, was relatively low. In 8 (34.8%) of 23 patients, central or paracentral ERG amplitudes were normal, whereas perimetric sensitivity was abnormally reduced. Poor sensitivity may suggest that postreceptoral structures in the proximal retina, not contributing to the FERG, are involved in the cone dysfunction of patients with RP. This is consistent with previous studies in human patients ^{32 45 46 47} and in animal models of RP, ^{48 49} showing that from early disease stages proximal retinal layers may be affected independently of, and additively to, the dysfunction of the distal retina. 

In summary, the results of this study show regional variations in the cone system in RP that can be evaluated by FERG 1F recordings. These alterations, involving both response amplitude and phase, tend to be TF dependent and to differ, depending on retinal location. The correlation between FERG amplitudes and perimetric sensitivity losses indicates that the FERG recorded at 41 Hz is highly specific in predicting normal field sensitivity, at both central and paracentral locations, in individual patients with RP. The data suggest the potential clinical use of the present FERG method, alone or in combination with other techniques, ¹¹ to probe and characterize local cone system dysfunction in RP. 

 Supported by Telethon Grant E833, Rome, Italy.

 Submitted for publication December 19, 2001; revised June 11, 2002; accepted July 19, 2002.

 Commercial relationships policy: N.

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

 

Table 1.

 

Clinical Findings in Patients

Patient	Age	Visual Acuity	Inheritance Mode	Ganzfeld Cone ERG (% loss)	Visual Field (°) (Goldmann)	SF*	Central Visual Field (10-2) Mean Sensitivity Loss (dB), †
														Central	Paracentral
1	21	20/25	SIMPLE	50	>40	2.4	16	22
2	59	20/20	SIMPLE	40	>50	1.8	4	16
3	64	20/30	SIMPLE	60	>50	3.6	6	11
4	31	20/20	SIMPLE	50	>30	2.5	4	6
5	56	20/20	SIMPLE	60	>20	0.6	10	25
6	36	20/60	SIMPLE	70	>30	3.4	5	10
7	39	20/50	AR	70	>30	3.1	4	10
8	24	20/100	AR	60	>50	1.3	5	9
9	46	20/100	SIMPLE	70	>50	2.5	6	10
10	37	20/30	USHER	70	>40	3.9	13	27
11	26	20/40	AD	70	>30	1.6	15	27
12	34	20/20	SIMPLE	50	>50	1.5	0	1
13	15	20/20	USHER	70	>30	3.0	6	30
14	38	20/20	USHER	70	>30	3.2	8	32
15	19	20/25	AD	40	>30	1.5	3	11
16	36	20/100	SIMPLE	60	>20	0.9	11	16
17	58	20/100	SIMPLE	60	>20	1.8	9	17
18	56	20/30	USHER	60	>30	1.3	4	13
19	31	20/50	SIMPLE	40	>20	2.4	9	17
20	23	20/30	AR	30	>30	2.1	17	20
21	45	20/20	SIMPLE	10	>30	1.2	10	21
22	40	20/20	SIMPLE	10	>40	1.8	8	10
23	25	20/25	AD	20	>30	1.2	4	9

Figure 1.

 

(A) Retinal location of central (diameter, 4.5°; eccentricity from the fovea, 2.25°) and paracentral (diameter 18°; eccentricity from the fovea, 4.5–9°) stimuli. (B) Representative examples of FERG responses obtained from a control subject with central and paracentral stimuli. Each record represents the final average of eight blocks of 200 events each. The Fourier-isolated 1F components of both responses are superimposed on the raw records. One full response cycle in the 24-ms epoch is shown for either central or paracentral response. On the right of each record, plots of cosine and sine values of each block average of 200 events are shown. Each data point represents a vector, the length of which reflects the response amplitude and the orientation, the phase angle. Phase delay is in a clockwise direction. (C) FERG responses from the same subject normalized by the stimulus area (i.e., response densities).

Figure 2.

 

Mean FERG amplitudes (±SEM), expressed as response densities, recorded at different TFs from normal subjects and patients in response to central (top) and paracentral (bottom) stimuli. Patient data are separately presented for the subgroups with visual acuity better (n = 16) or worse (n = 7) than 20/40.

Figure 3.

 

Average (± circular standard deviations) FERG phase values, recorded as a function of TF for central (A) and paracentral (B) stimuli in normal subjects and patients with RP. Patients’ data are shown as average results from the whole group (top) and separately for the subgroups with visual acuity better (n = 16) or worse (n = 7) than 20/40 (bottom).

Figure 4.

 

Cosine and sine values of (left) 10.3- and (right) 41-Hz 1F components, recorded from the central retinal region, plotted for normal subjects and patients together with the corresponding individual noise values.

Figure 5.

 

Perimetric sensitivity losses, recorded individually in patients with RP for central (A) and paracentral (B) stimuli, plotted as a function of the corresponding 41-Hz FERG response density losses. For both parameters, log data values are shown. The lower 95% normal confidence limits for response density and perimetric sensitivity are also shown.

Table 2.

 

Diagnostic Sensitivity, Specificity, and Accuracy of the Central and Paracentral ERG Amplitudes in Predicting Perimetric Sensitivity Losses at Corresponding Retinal Locations

	Reduced Perimetric Sensitivity	Normal Perimetric Sensitivity
Central*
Reduced ERG amplitude	7	1
Normal ERG amplitude	8	7
Paracentral
Reduced ERG amplitude	14	0
Normal ERG amplitude	8	1