Photopic Negative Response of the Human ERG: Losses Associated with Glaucomatous Damage

Abstract

purpose. To evaluate in glaucomatous eyes the photopic electroretinogram (ERG) negative response (PhNR), a component that follows the b-wave peak and is thought to be correlated with inner retinal activity.

methods. Eleven patients with open-angle glaucoma (OAG) and moderate field loss (Humphrey 30-2 [Humphrey Instruments, San Leandro, CA] mean deviation≤ −6 dB), eight with ocular hypertension (OHT), and eight age-matched normal subjects were tested. Optic discs of patients and control subjects were evaluated by confocal scanning laser ophthalmoscopy. ERGs were recorded to long-duration stimuli (250 msec) of photopic luminance (78 candelas [cd] /m²), presented in the macular region (12° × 12° field size) on a steady, adapting background. Amplitudes of the a-wave and b-wave and the PhNR were measured. Pattern reversal ERGs to 30-minute checkerboards were also recorded from patients and control subjects.

results. Compared with control subjects, patients with OAG showed reduced PhNR (average reduction: 62%, P < 0.01), but normal a- and b-wave amplitudes. In patients with OHT, PhNR and a- and b-wave amplitudes did not differ from control values. In individual patients with OAG, PhNR amplitudes were correlated positively with pattern ERG amplitudes (r = 0.80; P < 0.01) and central (12°) perimetric mean deviations (r = 0.68; P < 0.05) and negatively with cup-to-disc area ratios (r = −0.79; P < 0.01) and cup shape measures (r = −0.78; P < 0.01).

conclusions. Similar to that found in monkeys with experimentally induced glaucoma, the PhNR is selectively altered in human glaucoma. The correlation between PhNR losses and clinical parameter abnormalities suggests that this component depends on inner retina integrity and may be of clinical value for detecting glaucomatous damage.

The cone-mediated electroretinogram (ERG) in response to brief or long-duration flashes is the sum of component potentials reflecting the electrical activity of different retinal generators. ^{1 2 3} Photopic a-wave reflects cone photoreceptor activity, ^{4 5 6} with a contribution from hyperpolarizing postreceptoral cells. ² The b-wave is shaped by the activity of both depolarizing and hyperpolarizing cone bipolar cells. ³ Oscillatory potentials are thought to reflect mainly the activity of amacrine cells. ⁷ The d-wave of ERGs to long-duration stimuli reflects the contribution of both cone photoreceptors and hyperpolarizing bipolar cells. ⁸ Recently, a negative potential after the b-wave (and the d-wave in the responses to long-duration stimuli) has been described in the photopic ERG (photopic threshold response ⁹ ; photopic negative response [PhNR] ¹⁰ ). In the monkey, ¹⁰ this component has been found to be selectively affected by experimental glaucoma or intraocular injection of tetrodotoxin (TTX), a drug known to suppress action potentials of retinal ganglion cells and amacrine cells. These findings support the hypothesis that the PhNR of the primate cone ERG originates from the inner retina, directly from retinal ganglion cells, or perhaps through mediation by glia ^{10 11} and raises the possibility that this component, similar to the pattern electroretinogram (PERG), ^{12 13 14 15} could be useful to evaluate inner retinal function in human glaucoma. 

In the present study, PhNR of the cone-mediated ERG was evaluated in normal subjects and in patients with clinically defined open-angle glaucoma (OAG) or ocular hypertension (OHT). The purpose was to determine whether this component was altered in glaucoma and was correlated with the extent of glaucomatous damage assessed by standard diagnostic methods. For this purpose we compared, in the same affected eyes, the PhNR with the PERG, perimetric sensitivity, and optic disc morphology. The results showed that PhNR, unlike other ERG components, was significantly altered in glaucomatous eyes and that losses of this component were correlated with abnormalities of clinical parameters recorded from the same glaucomatous eyes. 

Methods

Subjects

Eleven patients (six men and five women, mean age: 49.3 ± 8.2, range: 36–62 years) with a diagnosis of OAG and eight patients (four men and four women, mean age: 48 ± 6.7, range: 40–57 years) with OHT were included in the study. Diagnosis of OAG was established on the basis of an elevated IOP (>21 mm Hg on two separate occasions), an open angle, the presence of abnormal white-on-white perimetry (Humphrey 30-2; Humphrey Instruments, San Leandro, CA) with a typical reproducible defect (arcuate and/or paracentral scotoma or nasal step; three or more adjacent points, not contiguous with the field borders, with a ≥5 dB loss, or two or more adjacent points, not contiguous with the field borders, with ≥10 dB loss) and glaucomatous optic disc, evaluated by slit lamp biomicroscopy and 78-D lens, with a cup-to-disc ratio greater than 0.6 (or an interocular cup-to-disc ratio asymmetry ≥0.2) and one or more of the following disc abnormalities: excavation, thinning of the rim, notching, nerve fiber layer defects, or parapapillary atrophy. Field loss was graded from early to moderate in all patients, with Humphrey 30-2 mean deviation ranging from −1.2 to −6 dB. Patients with OHT had an elevated IOP (>21 mm Hg on two separate occasions), normal Humphrey perimetry, and normal clinical optic disc appearance. All patients were selected by an experienced ophthalmologist (AC) specializing in glaucoma. Eight normal subjects, whose sex and age distribution was comparable with that of patients (four men and four women, mean age: 48.7 ± 6.4, range: 38–59), provided normative ERG values. In Table 1 , clinical findings obtained in individual patients with OHT and OAG are reported. In all patients, analysis of the optic disc was performed by confocal scanning laser ophthalmoscopy (Heidelberg Retina Tomograph, [HRT]; Heidelberg Engineering, Heidelberg, Germany) according to a previously published protocol. ¹⁶ Among the various morphometric parameters obtained by HRT, those most sensitive and specific for glaucoma damage ^{17 18 19} were considered in the analysis: the cup-to-disc area ratio and the cup shape measure. Cup-to-disc area ratio is the ratio of cup to disc surface determined by means of the current HRT software algorithm (2.01 ver.). The cup shape measure is a structural index summarizing in one number the depth variation and steepness of the cup walls. ¹⁸ Its value is typically negative in normal eyes and less negative or positive in glaucomatous eyes. ¹⁸ These measurements were compared with the 95% confidence limits established in 18 normal eyes (18 subjects; mean age: 44 ± 6.5 years, range: 38–59 years), with disc area in the ranges of 1 to 2 mm² (n = 9; mean cup-to-disc area ratio = 0.15, upper 95% confidence limit = 0.22; mean cup shape measure = −0.29, upper 95% confidence limit = −0.14) and 2 to 3 mm² (n = 9; mean cup-to-disc area ratio = 0.23, upper 95% confidence limit = 0.36; mean cup shape measure = −0.23, upper 95% confidence limit = −0.06). Subjects providing normative HRT and electroretinographic data belonged to independent groups. Cup-to-disc area ratio was significantly altered in 2 of 8 OHT eyes and in 7 of 11 OAG eyes, whereas cup shape measure was normal in all OHT eyes and abnormal in 2 of 11 OAG eyes. In Table 1 , in addition to the Humphrey mean deviation derived from the 30° field, a central mean deviation calculated from a field area of 12° × 12° is reported for each patient. This mean deviation value was obtained by averaging the local losses in decibels found in the total deviation plot of the Humphrey 30-2 and was used for correlation purposes with the electroretinographic responses, elicited from the same central 12° × 12° stimulation area (see below). Informed consent was obtained from every subject or patient after the procedures used in the study were fully explained. The research followed the tenets of the Declaration of Helsinki. 

ERG Recording

Stimuli were electronically generated on a computer monitor (definition: 320 × 200 pixels; frame rate: 70 Hz; mean luminance: 78 candelas [cd]/m²; modulation depth: 92%) that was masked with a white, opaque cardboard, leaving uncovered a square field of 12° × 12° (9 × 9 cm at the viewing distance of 43 cm; definition of the cutout: 119 × 74 pixels) in size. Uniform field on–off stimulation was obtained at a rate of 2 Hz, so that the field was on and off every 250 msec. The trigger signal was operated at the start of every on-phase of the stimulation. The stimulator was peripherally surrounded with a large (70 × 70 cm) white cardboard that was floodlit at 150 cd/m² (i.e., luminance of the bright stimulus phase) by means of a projector (Prado, Germany). Care was taken (by occluding the projector’s light beam using a field stop mounted on a slide) to avoid illumination of the stimulating area of the monitor. This procedure was effective in keeping unaltered the contrast on the screen, while minimizing the stray-light effects, so that the responses could be considered focal (focal ERG, FERG). ^{20 21 22} Subjects and patients looked at a small (5-minute) fixation spot placed in the center of the stimulating field. Pupils were natural, and their size was measured under adaptation at the stimulus mean luminance. No differences in pupil size were observed between control subjects and patients or between the two patient groups. 

ERGs were recorded by an Ag-AgCl electrode taped on the skin of the lower eyelid. A similar electrode, placed over the eyelid of the contralateral unstimulated eye, was used as reference (interocular recording). ^{13 23 24} Responses were amplified, filtered (0.3–100 Hz, −6 dB per octave), sampled at 2 kHz with a resolution of 12 bits, and averaged (300 events, three blocks of 100 events each) with automatic artifact rejection. Two recordings were obtained to verify reproducibility. Responses to on–off long-duration stimuli consisted of on and off components. Examples of typical FERG waveforms to such stimulation have been published. ²² The on component consisted of an a-wave, a b-wave, and a negative potential (PhNR) after the b-wave peak and peaking at 80 to 95 msec. The off component consisted of an early and a late positive component, peaking at 35 and 50 msec after the stimulus cessation, respectively, and a negative component. The latter was more variable in amplitude and peak time than the negative on potential. ¹⁰ This variability was related to a variation in the baseline after the termination of the on stimulus. Given the uncertainty of the measurement, the off response component was not further considered in the analysis. For each record, amplitude and time to peak of on components were measured. The following parameters were evaluated: a-wave amplitude, b-wave amplitude, and time to peak, and PhNR amplitude and time to peak. For each subject or patient, a noise response was also measured while the subject fixated an unmodulated field of the same mean luminance as the stimulus. Noise peak-to-peak amplitude was in all cases less than 0.15μ V, when considering a temporal window corresponding to that at which the response component was expected to peak. The signal-to-noise ratio for each component was measured by dividing the peak amplitude of that component by the noise in the corresponding temporal window. Responses were considered as recordable only if the signal-to-noise ratio was equal to or greater than 2.5. In all normal subjects and patients, each FERG amplitude component satisfied this signal-to-noise ratio criterion. 

In all normal subjects and patients, a PERG was also recorded according to a previously published technique, ²⁵ with full refractive correction for the test distance (43 cm). Stimulation field size was a square subtending, the same as for the focal ERG, 12° × 12°. Checkerboards of 30 minutes of angular subtense (mean luminance: 78 cd/m²; contrast: 92%), modulated in counterphase at four reversals per second, were used as a stimuli. The amplitude of the postive–negative complex P₅₀–N₉₅ was measured. 

Statistical Analysis

Although both eyes for each subject or patient were tested, only the results from the right eyes were included in the statistical analyses. Results from normal subjects and patients were analyzed by one-way analysis of variance (ANOVA) with post hoc Tukey honestly significant difference (HSD) tests for multiple comparisons. Pearson’s correlation and linear correlation analysis was used to correlate amplitudes of the various FERG components with the corresponding clinical parameters (pattern ERG, central perimetric mean deviation, cup-to-disc area ratio, cup shape measure) recorded in the individual patients with OAG. In all the analyses, P < 0.05 was considered significant. 

Results

Figure 1 shows the FERGs (two recordings) and the clinical findings (i.e., disc appearance by HRT and visual field by Humphrey 30-2) obtained in a normal subject, a patient with OHT and a patient with OAG. Figure 1A , top, also shows how the a-wave, b-wave, and PhNR of the on response were measured. In the patient with OHT, b-wave and PhNR were similar in both amplitude and time to peak to those of the control subject. The a-wave was moderately reduced in amplitude. In the patient with OAG, the a- and b-wave amplitudes and peak times were similar to control values, whereas the PhNR was selectively reduced in amplitude, compared with either the control or the OHT response. PhNR peak time of the OAG response was comparable with normal. 

In Figure 2 , the frequency distribution histograms of the a-wave, b-wave, and PhNR amplitudes are reported for the different groups of the study population. It can be seen that the distributions of the three groups for all three amplitude measures overlap. However, the distribution of PhNR amplitude, unlike those of a- and b-waves, is substantially shifted toward lower values in the OAG group, compared with both control and OHT groups. Mean (±SE) amplitudes and peak times of the different ERG components, obtained from the three groups of the study population, are reported in Table 2 . ANOVA showed that, on average, a-wave and b-wave amplitudes and b-wave and PhNR times to peak did not change significantly, whereas PhNR amplitude showed a significant variation across groups (F_2,24 = 10.34; P < 0.001). The Tukey tests showed significant differences in PhNR mean amplitude between control subjects and patients with OAG (62%; P < 0.01) as well as between patients with OHT and those with OAG (53%; P < 0.05). Mean PERG amplitude changed significantly across groups (F_2,24 = 10.15; P < 0.001). The Tukey tests showed significant differences in mean amplitude between control subjects and patients with OHT (37%; P < 0.05), as well as between control subjects and patients with OAG (54.3%; P < 0.01). 

Figure 3 shows scatterplots of PhNR amplitudes, recorded from individual patients with OAG, as a function of the corresponding values of pattern ERG amplitude (Fig. 3A) , perimetric mean deviation (Fig. 3B , average value within the central 12° of visual field), cup-to-disc area ratio (Fig. 3C) , and cup shape measure (Fig. 3D) . There were significant correlations between PhNR amplitudes and all the parameters shown (r values and corresponding slopes of linear regressions, fitted to the data points, are reported in the figure). PhNR amplitudes were positively correlated with pattern ERG amplitudes (P < 0.01) and perimetric central mean deviations (P < 0.05). In addition, PhNR amplitudes were negatively correlated with cup-to-disc area ratios (P < 0.01) and cup shape measures (P < 0.01). Pattern ERG amplitudes of OAG eyes were significantly correlated (not shown) with perimetric mean deviations (r = 0.6, P < 0.05) and cup shape measures (r = −0.74, P < 0.01). In contrast to the FERG PhNR component and the PERG, neither a- or b-wave amplitudes was significantly correlated with any of the clinical parameters included in the analysis. 

Discussion

The results of this study show that the PhNR component of the focal cone ERG was, on average, significantly reduced in amplitude in OAG eyes, compared with the responses obtained from either normal control or OHT eyes. This reduction appeared to be specific for PhNR, because the mean peak amplitudes of the a- and b-waves did not differ significantly between control subjects and both groups of affected eyes. In the normal control group, interindividual variabilities of a- and b-wave amplitudes were larger than those of PhNR amplitude (coefficients of variation of 60% and 48% for a- and b-wave amplitudes, respectively, compared with 37% for PhNR amplitude). This difference in variability, which has to be confirmed by further studies of larger subjects samples, may have obscured to some extent the differences in mean a- and b-wave amplitudes across the groups of the study population. However, the average reduction in PhNR amplitude found in OAG eyes, compared with that in either normal or OHT eyes, was far greater than corresponding losses in a- and b-waves (Table 2) , indicating a predominant, or even selective, effect of disease on the FERG PhNR component. The present findings are in agreement with those previously reported in monkeys, ¹⁰ showing that PhNR was selectively reduced in eyes with experimentally induced glaucoma with respect to contralateral, untreated eyes. 

The cellular origin of the PhNR component has not been clearly established. However, the available evidence in cats and primates suggests that this component originates from the inner retina. In the cat, ¹¹ intraretinal recordings showed that TTX-sensitive negative potentials, with a time course similar to the PhNR of the ERG, are present in the proximal retina of area centralis, where ganglion cell density is high. In addition, these components are particularly pronounced at recording sites approaching the optic nerve head. Further evidence in cats ¹¹ also suggests a glial mediation of the TTX-sensitive PhNR of the photopic ERG. In the monkey, intravitreally injected TTX ¹⁰ abolishes the PhNR selectively, leaving the other ERG components relatively unaltered. Because TTX is a voltage-gated Na⁺ blocker, and the retinal ganglion cells, their axons, and the amacrine and interplexiform cells are the major retinal elements that possess voltage-gated Na⁺ channels, ^{26 27} these neurons would be, as suggested previously, ¹⁰ those most likely responsible for PhNR generation. In humans, the spatial distribution of the focal ERG PhNR was previously investigated in our laboratory. ²² In that study, the negative ERG component after the b-wave peak was identified as a PIII component and was thought to be linked to the activity of outer retina. Indeed, no experimental data were available at the time the study was conducted to support an inner retinal origin for PhNR. Of note, PhNR amplitude showed a maximum response density in the foveal region, where ganglion cell density is highest, and its distribution differed from that of the b-wave, whose response density was much less pronounced in the same foveal area. ²² These results in humans, indicating that PhNR spatial distribution may follow that of retinal ganglion cells, are consistent with the experimental data in cats and lend further support to the hypothesis that the human PhNR is generated from inner retina. 

In this study, unlike that in monkeys by Viswanathan et al., ¹⁰ ERGs were elicited by stimuli restricted to the macular region on a light-adapting background. A focal instead of a ganzfeld stimulation was chosen for two reasons. One reason was that the results of previous studies in normal subjects or in monkeys ^{22 28} indicated that the PhNR is relatively more represented and less masked by the b-wave in the focal compared with the full-field cone ERGs. The other reason was that we sought to compare in glaucomatous eyes PhNR, perimetric sensitivity, and PERG recorded from comparable retinal areas in the macular region. We found that losses in PhNR amplitude observed in OAG eyes were significantly correlated with perimetric sensitivity losses. A similar, although weaker, correlation between PhNR amplitude and behaviorally determined perimetric losses has been also reported in glaucomatous monkeys. ¹⁰ The present results in humans and the monkey data suggest that PhNR and perimetric losses in glaucoma share, at least in part, a common mechanism presumably related to ganglion cell dysfunction. In our patients with OAG, a significant correlation was found between PhNR and PERG P₅₀-N₉₅ amplitude losses. This suggests that PhNR may be directly, or indirectly, related to ganglion cell activity, because the PERG N₉₅ is thought to reflect specifically functional integrity of these neurons (see, for example, References 29, 30). It should be noted, however, that PERG amplitude was, on average, comparatively more affected than PhNR in OHT eyes (see post hoc multiple comparisons and Table 2 ), indicating a greater vulnerability for the PERG to subclinical glaucomatous damage. PhNR losses of individual patients with OAG were also found to be correlated with the severity of optic disc cupping as expressed by two morphometric parameters, the cup-to-disc area ratio, and the cup shape measure. There is histologic and clinical evidence that disc cupping abnormalities may be an indirect sign of neural damage in glaucoma. ^{16 17 18 31} Recent clinical studies ^{16 17 18} indicate that early glaucomatous neural dysfunction can be accurately predicted by the value of cup shape measure, an index of depth variation and steepness of the cup walls. ¹⁸ The association found in this study between PhNR and disc morphometric parameters, particularly the cup shape measure, support the hypothesis that PhNR losses may represent a specific indicator of glaucomatous damage. Clearly, further cross-sectional and longitudinal studies, in a larger sample of patients, are needed before the diagnostic accuracy and predictive value of PhNR in glaucoma can be fully established. 

Compared with PERG, recording of cone ERG PhNR may present some practical advantages. PhNR recording does not require full refractive correction for the test distance, which is otherwise critical for PERG measurement. Furthermore, the response loss should be specific for glaucomatous damage also in eyes with moderate optical media opacities, which preclude recording of the PERG. Indeed, PhNR response losses in glaucomatous monkey eyes did not show a dependence on either mean luminance or modulation depth. ¹⁰ Therefore, it is presumable that PhNR, unlike the PERG, may not be significantly affected in nonglaucomatous eyes with early or moderate media opacities. These technical advantages, in addition to the presumed retinal origin, further point to a potential clinical use of the PhNR as a tool in clinical detection of glaucomatous damage. 

 Submitted for publication October 21, 1999; revised January 28, 2000; accepted February 10, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Demographic and Clinical Findings in Study Patients

Patient/Sex/Age (y)	Diagnosis	IOP (mm Hg)	Humphrey Mean Deviation (dB; 30°)	Humphrey Mean Deviation (dB; 12°)	HRT Disc Area (mm2)	HRT Cup to Disc Area Ratio	HRT Cup Shape Measure
7/M/57	OHT	25	0.03	0.05	1.74	0.17	−0.20
9/F/52	OHT	24	0.01	0.02	1.84	0.25*	−0.18
10/F/40	OHT	24	0.50	0.45	1.87	0.15	−0.25
14/M/48	OHT	24	−0.65	−0.75	2.40	0.34	−0.15
17/F/56	OHT	23	−0.20	−1.62	1.97	0.11	−0.28
18/M/48	OHT	25	0.11	0.20	2.92	0.27	−0.17
24/M/43	OHT	24	−0.42	−0.88	2.15	0.21	−0.13
27/F/40	OHT	25	−0.73	−0.93	1.82	0.34*	−0.24
1/M/36	OAG	26	−3.10	−0.94	2.28	0.41*	−0.14
11/M/50	OAG	28	−1.20	−1.63	1.54	0.25*	−0.19
12/M/50	OAG	26	−5.00	−1.00	2.09	0.02	−0.31
13/M/47	OAG	28	−2.81	0.06	1.75	0.08	−0.20
15/M/45	OAG	27	−4.00	−1.50	2.12	0.19	−0.19
16/M/62	OAG	32	−6.00	−4.00	2.40	0.55*	−0.10
19/F/47	OAG	28	−4.50	−3.80	2.30	0.28	−0.17
21/F/60	OAG	24	−3.20	−1.88	2.95	0.40*	−0.13
22/F/46	OAG	24	−1.38	−2.19	1.45	0.26*	−0.22
25/F/40	OAG	28	−5.87	−4.13	2.41	0.62*	−0.01*
26/F/59	OAG	26	−3.48	−3.06	2.94	0.43*	−0.05*

*  Above normal 95% confidence limits.

Figure 1.

 

(A) FERG responses to on–off long-duration stimuli obtained from a normal subject, a patient with OHT (patient 27 in Table 1 ) and one with OAG (patient 25 in Table 1 ). Arrows on the top trace indicate how the a-wave, b-wave, and PhNR of the on response are measured. Dotted line: baseline level. (B) Optic disc appearance by retina tomograph reflectance images. (C) Automated visual fields (total deviation plots from the perimetry threshold test). Disc images and fields obtained from the same eyes for which FERGs are reported in (A). D area, disc area in square millimeters; C/D area, cup-to-disc area ratio; CSM, cup shape measure; MD, mean deviation in dB.

Figure 2.

 

Frequency distribution histograms of the a-wave, b-wave, and PhNR amplitudes recorded from the study groups.

Table 2.

 

Electrophysiological Results Obtained in Normal Subjects and Patients

	Normal (n = 8)	OHT (n = 8)	OAG (n = 11)
Focal ERG
a-Wave amplitude (μV)	0.89 ± 0.19	0.75 ± 0.15	0.685 ± 0.09
b-Wave amplitude	1.64 ± 0.28	1.14 ± 0.16	1.22 ± 0.13
b-Wave time to peak (msec)	46 ± 1.4	49 ± 1.3	48 ± 1.4
PhNR amplitude	1.92 ± 0.26	1.54 ± 0.22	0.73 ± 0.12, † , ‡
PhNR time to peak	89.6 ± 3.1	92.2 ± 1.3	86.2 ± 1.9
Pattern ERG
P50–N95 amplitude	2.54 ± 0.18	1.59 ± 0.17*	1.16 ± 0.25, †

*  Significantly different from the normal mean value (Tukey HSD) at P < 0.05.

†  Significantly different from the normal mean value (Tukey HSD) at P < 0.01.

‡  Significantly different from the OHT mean value (Tukey HSD) at P < 0.05.

Figure 3.

 

PhNR amplitudes, recorded from individual patients with OAG, plotted as a function of the corresponding values of the four study parameters. r, regression coefficient; B, slope of the regression line.* P < 0.05; **P < 0.01.