Heterochromatic Flicker Electroretinograms ReflectingLuminance and Cone Opponent Activity inGlaucoma Patients

Mirella Telles Salgueiro Barboni,1,2 Gobinda Pangeni,1 Dora Fix Ventura,2 Folkert Horn,1

and Jan Kremers1

PURPOSE. To measure heterochromatic flicker electroretino-grams (ERGs) at high (36 Hz) and intermediate (12 Hz) tem-poral frequencies to evaluate luminance and cone opponentresponses, respectively, in glaucoma eyes with (perimetric)and without (preperimetric) visual field defects.

METHODS. Flicker ERGs were recorded from one randomlychosen dilated eye of 32 patients (mean age, 61 � 11 years; 15men, 17 women) from the Erlangen Glaucoma Registry andfrom 24 healthy volunteers (mean age, 43 � 11 years; 14 men,10 women). Red and green light-emitting diodes in a Ganzfeldstimulator were sine wave-modulated in counterphase. Theresponses were measured at 36 Hz, the frequency at whichERGs reflect activity of the luminance pathway, and at 12 Hz,the frequency at which ERGs reflect chromatic activity.

RESULTS. Response amplitudes were similar in glaucoma pa-tients and controls. Phase differences were observed in pa-tients with visual field defects (perimetric) compared with thecontrol group at 36 and 12 Hz in the first harmonic and secondharmonic responses. Patients without visual field defects (pre-perimetric) showed phase differences for the second harmoniccomponent at 36 Hz. No age effect on response amplitudes andphases was found in any of the subject groups (controls andpatients).

CONCLUSIONS. The responses displayed phase differences butnot amplitude differences in perimetric glaucoma patients atboth 36 and 12 Hz, suggesting that both magnocellular andparvocellular pathways are affected. Preperimetric glaucomapatients also showed phase differences. The response phasemay be sensitive to early dysfunction of the inner retina. (Clini-calTrials.gov number, NCT00494923.) (Invest Ophthalmol VisSci. 2011;52:6757–6765) DOI:10.1167/iovs.11-7538

Glaucomatous degeneration of retinal ganglion cells is ofteninduced by elevated intraocular pressure (IOP), leading to

progressive visual loss.1 Glaucoma may primarily affect largeroptic nerve fibers.2,3 In contrast, experimental glaucoma innonhuman primates has been shown recently to lead to pro-gressive nonselective neural impairment of visual pathwaysand dysfunction of both large and small retinal ganglion cellsand their magnocellular (M) and parvocellular (P) geniculocor-tical projections.4–6 Using a psychophysical paradigm de-signed to selectively detect the function of the M and P visualpathways,7 glaucomatous visual dysfunction was indeed foundin both pathways.8–10

Several electrophysiological11–13 and psychophysical14–17

studies have assessed the activity of specific visual pathways inglaucoma patients.18–20 One noninvasive electrophysiologicaltool that is available to study the functional integrity of thehuman retina is the electroretinogram (ERG), which measuresa complex electrical mass response of the retina recorded atthe cornea and generated by different cell groups.21 Manystudies sought to elucidate the cellular origins of the flashERG,22,23 flicker ERG,24,25 and pattern ERG.26

Flicker ERG responses appear to be mainly generated bypostreceptoral retinal activity.27,28 At temporal frequenciesabove approximately 30 Hz, the flicker ERG is dominated bythe fundamental Fourier component (first harmonic),29 al-though higher harmonic components that are thought to begenerated in the inner retina may sometimes be present.24 Thedifferent harmonic components may reflect the activity ofdistinct retinal cells.30 At intermediate temporal frequencies(between 10 and 20 Hz), in addition to the fundamental com-ponent, there is a substantial second harmonic contribution toluminance photopic flicker ERG responses.29 Photoreceptorsappear to make a larger contribution to the first harmoniccomponent,31 and the second harmonic component may rep-resent the spiking activity of the inner retina.25

Although postreceptoral mechanisms strongly influenceflicker ERG responses, correlating the ERG response with Mand P activity in the retinogeniculate pathways is difficult.Flicker ERGs measured with heterochromatic stimuli at hightemporal frequencies (above approximately 20 Hz) displayproperties that are similar to those of the magnocellularlybased psychophysical luminance channel.32–35 Recently, workfrom our laboratory provided evidence that the high-frequencyflicker ERG probably directly reflects M activity. Additionally,under specific conditions (stimulus frequency of approxi-mately 12 Hz and the presence of a red-green chromatic com-ponent), the first harmonic component of flicker ERG re-sponses reflects cone opponent pathway activity.36,37 Theresults of these studies suggest that objectively investigatingthe responses of the M and P pathways in glaucoma patients ispossible. Furthermore, electrophysiological measurements of-fer the ability to study not only the amplitudes but also the

From the 1Department of Ophthalmology, University HospitalErlangen, Erlangen, Germany; and the 2Neuroscience and BehaviorCenter and Department of Experimental Psychology, University of SaoPaulo, Sao Paulo, Brazil.

Presented in part at the annual meeting of the Association forResearch in Vision and Ophthalmology, Fort Lauderdale, Florida, May2010.

Supported by German Research Council Grant KR1317/9-1. MTSBreceived an FAPESP fellowship for graduate students (07/55125-1).DFV is a CNPq Research Fellow. JK is Fellow in the Excellence Programof the Hertie Foundation.

Submitted for publication March 10, 2011; revised May 12, 2011;accepted June 3, 2011.

Disclosure: M.T.S. Barboni, None; G. Pangeni, None; D.F. Ven-tura, None; F. Horn, None; J. Kremers, None

Corresponding author: Jan Kremers, Department of Ophthalmol-ogy, University of Erlangen-Nuremberg, Schwabachanlage 6, 91054Erlangen, Germany; jan.kremers@uk-erlangen.de.

Clinical Trials

Investigative Ophthalmology & Visual Science, August 2011, Vol. 52, No. 9Copyright 2011 The Association for Research in Vision and Ophthalmology, Inc. 6757

phases of the responses, which is not possible with psycho-physical techniques. Based on these results, the present studydetermined how early postreceptoral retinogeniculate path-ways are affected by glaucoma.

SUBJECTS AND METHODS

Subjects

Measurements were performed in one randomly chosen eye of 32glaucoma patients (mean age, 61 � 11 years; 15 men, 17 women) fromthe Erlangen Glaucoma Registry. This group of patients was dividedinto two subgroups, those with perimetric open-angle glaucoma (n �20; mean age, 63 � 11 years; 10 men, 10 women) and those withpreperimetric open-angle glaucoma (n � 12; mean age, 59 � 11 years;5 men, 7 women) in accordance with the criteria previously estab-lished for conventional white-on-white automated perimetry, visualacuity, and IOP.38 Briefly, the inclusion criteria were open chamberangle, increased IOP (measurements �21 mm Hg), and abnormalappearance of the optic nerve head. Additionally, all perimetric glau-coma patients displayed glaucomatous visual field defects. Mean devi-ations ranged from 1.2 to 23.8 (average, 9.6 � 6.9) in the perimetricgroup and from �2.2 to 2.1 (average, 0.2 � 1.4) in the preperimetricgroup. The control group consisted of 24 healthy volunteers (meanage, 43 � 11 years; 14 men, 10 women).

All glaucoma patients (and the control subjects) were phakic andhad clear lenses. Any eye disorder (such as cataract) other than glau-coma was an exclusion criterion. In addition, normal color vision wasestablished. All participants underwent anomaloscope (HMC Anoma-loskop; Oculus, Wetzlar, Germany) examination before the ERG mea-surements. The anomalous quotient (AQ) was within the normal range(0.7–1.4), according to the manufacturer’s instruction manual. Perim-etric patients had AQs from 0.9 to 1.4 (average, 1.19 � 0.14), and

preperimetric patients had AQs from 0.8 to 1.2 (average, 1.02 � 0.17).Controls had AQs from 0.7 to 1.2 (average, 0.91 � 0.03). The exper-iments adhered to the tenets of the Declaration of Helsinki and wereapproved by the institutional ethics committee. Signed informed con-sent was obtained from the subjects after explanation of the nature andpossible consequences of the study.

Visual Stimulation

The stimuli were described previously.37,39 Briefly, a Ganzfeld stimu-lator (Q450 SC; Roland-Consult, Bradenburg, Germany) with six differ-ently colored light-emitting diodes (LEDs) was used. Only the red (638nm; bandwidth at half height, 19 nm; CIE coordinates: x, 0.6957; y,0.2966) and green (523 nm; bandwidth at half height, 36 nm; CIEcoordinates: x, 0.2016; y, 0.7371) LED arrays were used in the presentexperiment. The mean luminance of the red and green LED arrays was100 cd/m2 each, resulting in a mean retinal illuminance of approxi-mately 104 td (assuming a 8-mm pupil diameter) with a yellow meanchromaticity of 0.5813 (x) and 0.4030 (y) (CIE 1931 coordinates).

Figure 1A shows a sketch of the luminance outputs of the red andgreen LED arrays. Three different stimulus conditions were used, inwhich the fraction of red modulation depth was varied. At R/(R�G) �0, only the green LEDs were modulated with 100% contrast, whereasthe red LEDs were not modulated and had a constant output of 100cd/m2. At the R/(R�G) � 0.5 condition, the red and green LEDs weremodulated in counterphase with 50% contrast. R/(R�G) � 1 indicatesthe condition in which only the red LEDs were modulated with 100%contrast; the output of the green LEDs was 100 cd/m2 and unmodu-lated. The measurements were repeated at two temporal frequencies,36 Hz and 12 Hz.

Figure 1B displays the luminance (left) and red-green color oppo-nent (right) output of the 12-Hz stimuli. When R/(R�G) � 0 and 1, theluminance modulation was large and of equal amplitude. At

FIGURE 1. (A) Description of thestimuli. Red and green LEDs weremodulated in counterphase in a Gan-zfeld stimulator with a 100 cd/m2

mean luminance of each LED. Threedifferent stimulus conditions withdifferent fractions of red modulat-ion (red/[red�green]) were used:R/(R�G) � 0.0 (only the green LEDis modulated, whereas the red LED isconstant at 100 cd/m2), R/(R�G) �0.5 (red and green LEDs are modu-lated simultaneously in counter-phase), and R/(R�G) � 1 (only thered LED is modulated; the green LEDis constant at 100 cd/m2). Conditionswere presented at two temporal fre-quencies (36 and 12 Hz). (B, left)Resultant luminance modulation ob-tained by the addition of the red andgreen LED outputs. Time-averaged lu-minance output was 200 cd/m2 forall stimulus conditions. The output ofthe cone opponent system (right)was obtained by the subtraction ofthe red and green LED outputs. Thissubtraction resulted in identical sinu-soidal cone opponent modulationsfor all conditions. Furthermore, coneopponent modulation had the samephase in all conditions. Reprintedwith permission from Barboni MT,Ventura DF, Kremers J. Absence ofocular interaction in flicker ERG re-sponses reflecting cone opponentand luminance signals. Doc Ophthal-mol. 2010;121:69–75. © Springer.

6758 Barboni et al. IOVS, August 2011, Vol. 52, No. 9

R/(R�G) � 0.5, no luminance was output (note that the luminanceoutput was based on a luminance system with a spectral sensitivityaccording to the V�

40; different persons may display deviating spectralsensitivities so that the luminance output may differ slightly in them).The luminance modulation followed the green LEDs at R/(R�G) � 0and red LEDs at R/(R�G) � 1, so a 180° phase luminance shiftoccurred between the two conditions. Not only the amplitude but alsothe phase of the cone opponent system’s output was the same in allthree conditions because the stimulus was redder in the first half of astimulus period and greener in the second half under all conditions.The sinusoidal shape of the luminance output and cone opponentsystem had constant time-averaged values, indicating that the state ofadaptation was the same for all stimulus conditions (assuming that the36-Hz and 12-Hz stimulus frequencies were too high for adaptationprocesses). See Kremers et al.37 for details regarding a protocol thatallows the measurement of different R/(R�G) ratios and Barboni etal.39 for details regarding the short version of the protocol that wasused in the present work. The short protocol does not allow the studyof individual equiluminant points, which may vary. On the other hand,one intermediate condition was considered useful to verify whetherthe response amplitudes and phases followed the characteristic re-sponses previously reported,37 for a large population of healthy sub-jects in the present study. In addition, this short protocol, takingapproximately 5 minutes of measurement, can be easily applied inpatients.

Figure 2 shows the estimated response amplitudes (Fig. 2A, inarbitrary units) and phases (Fig. 2B, in degrees) of luminance (black)and red-green cone opponent (blue) mechanisms as a function of thestimulus condition. The S-cone input is assumed to be negligible, andthe total mean luminance of 200 cd/m2 is assumed to be too high forsubstantial rod input. The absence of an S-cone and a rod-driven signal

was shown previously by performing the measurements in a dichro-matic subject.37 Subsequently, we confirmed these data on other di-chromats (unpublished data, 2010). Observe that the estimated re-sponse amplitudes and phases follow the stimulus luminance and coneopponent modulation, respectively (Fig. 1B).

ERG Recordings

One eye was dilated with a drop of mydriaticum (0.5% tropicamide),and corneal ERG responses were measured with a DTL fiber electrodeattached from the outer to inner canthus of the eye. The reference andground skin electrodes were attached to the ipsilateral temple andforehead, respectively. The signals were amplified 100,000�, filteredbetween 1 and 300 Hz, and sampled at 1024 Hz using an electroreti-nography system (RetiPort; Roland Consult). At least 24 (36-Hz) and 48(12-Hz) 1-second episodes were averaged to obtain a large signal-to-noise ratio at both temporal frequencies.

Statistical Analysis

The recordings were Fourier analyzed using self-written software (Mat-Lab; The MathWorks, Natick, MA). The amplitudes (in microvolts) andphases (in degrees) of the first (fundamental) and second harmoniccomponents were statistically compared using analysis of variance(ANOVA) with subsequent multiple comparisons (including a Bonfer-roni post hoc correction for multiple testing) between controls andglaucoma patients. Pearson correlation coefficients were used to verifythe age effect. Noise was quantified by the average of the responseamplitudes of the 35- and 37-Hz components and the 11- and 13-Hzcomponents for the 36- and 12-Hz stimulus conditions, respectively.Phase values were disregarded when the signal-to-noise ratio was �3.

RESULTS

Figure 3 shows individual recordings of one control, one peri-metric, and one preperimetric patient at 36 Hz (upper record-ings) and 12 Hz (lower recordings). Consistent with previousdata,37,39 the 36-Hz response amplitudes of all subjects de-pended strongly on the R/(R�G) values. When R/(R�G) � 0and 1, the responses were large, whereas they were small inthe condition R/(R�G) � 0.5. Close inspection shows that theresponses at R/(R�G) � 0 and 1 were in counterphase witheach other. The 12-Hz responses displayed relatively constantamplitudes and phases for the different stimulus conditions.However, substantial frequency doubling was observed in allsubjects, especially for the condition R/(R�G) � 1.

Figure 4 shows the averaged (� SE) response amplitudesusing 36-Hz (left) and 12-Hz (right) stimuli measured in perim-etric patients (circles), preperimetric patients (triangles), andcontrols (squares). The upper plots show the first harmonicdata, and the lower plots show the second harmonic data as afunction of red fraction (R/[R�G]). Figure 5 displays the cor-responding response phases. A single asterisk indicates statis-tically significant differences (P � 0.05, Bonferroni correctionfor multiple tests) compared with controls, and double aster-isks indicate statistically significant differences compared withthe other two groups.

Consistent with previous data,37 the first harmonic re-sponses at 36 and 12 Hz followed the luminance and coneopponent output of the stimuli, respectively (Fig. 2, uppergraph). Accordingly, the three groups displayed a minimal firstharmonic response amplitude in condition R/(R�G) � 0.5 at36 Hz and did not vary with stimulus condition at 12 Hz (Fig.4, upper graphs). These data support the possibility of assess-ing different visual mechanisms using heterochromaticallymodulating stimuli, as previously proposed.37 At 36 Hz, thefirst harmonic response amplitudes were similar for bothgroups of glaucoma patients and the controls at all stimulusconditions. At 12 Hz, the first harmonic amplitude was larger

FIGURE 2. Calculated response amplitudes (A, arbitrary units) andphases (B, degrees) of the luminance (black) and red-green coneopponent (blue) output as a function of the stimulus condition.

IOVS, August 2011, Vol. 52, No. 9 Heterochromatic Flicker ERGs in Glaucoma Patients 6759

for perimetric patients in condition R/(R�G) � 0, but thedifference was not significant (P � 0.09).

The second harmonic response amplitudes at 36 Hz werelarger in perimetric patients in condition R/(R�G) � 1 than incontrols, but the difference was also not significant (P � 0.08).The second harmonic amplitudes were similar for the threegroups at 12 Hz for all conditions. The second harmoniccomponent at 12 Hz showed a similar dependency on R/(R�G)as the first harmonic component at 36 Hz; the responses weresmall for R/(R�G) � 0.5 and large for R/(R�G) � 0 and 1, withthe response larger for the red fraction of R/(R�G) � 1 thanR/(R�G) � 0 values (Fig. 4, lower graphs). This might indicatea residual input from a luminance mechanism at 12 Hz.

Phase data are shown in Figure 5. At 36 Hz, a 180° differ-ence was observed between the first harmonic phases in con-ditions R/(R�G) � 0 and R/(R�G) � 1 for the three groups. Incontrast, at 12 Hz, the first harmonic response phases wererelatively constant for patients and controls (Fig. 5, uppergraphs; note the different scales of the y-axes for the 36-Hz and12-Hz data). This is consistent with the proposal that lumi-nance and cone opponent mechanisms determine the re-sponses at 36 and 12 Hz, respectively37 (see also Fig. 2, lowergraph).

Phase differences between patients and controls werefound at 36 and 12 Hz (Fig. 5). The first-harmonic responsephases at 36 Hz were higher for the controls, indicating aresponse delay in the patients compared with controls, butonly in condition R/(R�G) � 1 was the difference betweenperimetric patients and controls significant (P � 0.0004). Thesecond harmonic response phases at 36 Hz in conditionsR/(R�G) � 0 and 1 were different between perimetric patientsand controls (P � 0.03 and P � 0.001, respectively) and also

between preperimetric patients and controls (P � 0.04 andP � 0.03, respectively).

The estimated time delay of the 36-Hz first harmonic re-sponses was 2 ms for the perimetric patients compared withthe controls (there was no significant difference between pe-rimetric and preperimetric glaucoma patients). For the 36-Hzsecond harmonic responses, the estimated time delays be-tween glaucoma patients (both perimetric and preperimetric)and controls were approximately 1.8 and 3 ms for theR/(R�G) � 0 and R/(R�G) � 1 conditions, respectively (as-suming that the delay is less than one period). Although thesetime delays are small, the phases can be measured very reliablyso that very small delay changes can be detected.

At 12 Hz, the first and second harmonic response phases weredifferent between perimetric patients and the other two groups incondition R/(R�G) � 0. The first harmonic response phases werehigher for the controls (P � 0.0001) and preperimetric patients (P �0.01) than for the perimetric patients. The second harmonic re-sponse phases were lower for the controls (P � 0.0008) and prep-erimetric patients (P � 0.01) than for the perimetric patients.

The estimated time delay for the perimetric patients wasapproximately 10 and 5.5 ms compared with the preperimetricpatients and the controls at 12 Hz in the first and secondharmonic components respectively.

To verify the relationship between visual field losses andERG response phases in glaucoma patients, response phaseswere plotted as a function of mean deviation values in perim-etric (circles) and preperimetric (triangles) glaucoma (Fig. 6)patients. The parameters used for this comparison were thefirst and the second harmonic response phases for conditionR/(R�G) � 0, in which phases are significantly different be-tween perimetric and preperimetric patients. There were sig-

FIGURE 3. Individual recordings of one control, one preperimetric patient, and one perimetric patient for the three conditions tested at 36 Hz (upper) and 12Hz (lower). The epoch shown is 250 ms. At 36 Hz, the response amplitudes changed strongly with R/(R�G) values. Responses at R/(R�G) � 0 and R/(R�G) �1 display larger amplitudes and are in counterphase to each other. At 12 Hz, the responses display relatively constant amplitudes and phases for the differentconditions.

6760 Barboni et al. IOVS, August 2011, Vol. 52, No. 9

nificant correlations between mean deviation values and re-sponse phases for both first (P � 0.01 and r2 � 0.25) andsecond (P � 0.04 and r2 � 0.15) harmonic response phases.

The average age of the control group (43 � 11 years) in thepresent study was 18 years younger than the average age of the

glaucoma patients (61 � 11 years). To verify whether thephase differences between patients and controls were notattributable to differences in age, response phases were plot-ted as a function of age. Figures 7 and 8 show individual first(upper graphs) and second (lower graphs) harmonic response

FIGURE 4. Averaged (� SE) re-sponse amplitudes at 36 Hz (left)and 12 Hz (right) in perimetric pa-tients (closed circles), preperimet-ric patients (closed triangles), andcontrols (open squares). Upper:first harmonic amplitudes; lower:second harmonic amplitudes. Nosignificant differences were ob-served (36 Hz, P � 0.08; 12 Hz, P �0.09; ANOVA followed by Bonfer-roni post hoc correction).

FIGURE 5. Averaged (� SE) re-sponse phases at 36 Hz (left) and 12Hz (right) temporal frequencies inperimetric patients (closed circles),preperimetric patients (closed trian-gles), and controls (open squares).Upper: first harmonic phases; lower:second harmonic phases. *Statisti-cally significant difference comparedwith controls only. **Statistically sig-nificant difference compared withthe other two groups. A significantphase change was observed in peri-metric patients compared with thecontrol group for the first harmonicresponse in condition R/(R�G) � 1at 36 Hz (P � 0.0004). The secondharmonic response phases at 36 Hzin conditions R/(R�G) � 0 and 1 aredifferent between perimetric pa-tients and controls (P � 0.03 and P �0.001, respectively) and also be-tween preperimetric patients andcontrols (P � 0.04 and P � 0.03,respectively). At 12 Hz, the first har-monic response phases were higherfor controls (P � 0.0001) and prep-erimetric patients (P � 0.01) than forperimetric patients, and the secondharmonic response phases werelower for controls (P � 0.0008) andpreperimetric patients (P � 0.01)than for perimetric patients.

IOVS, August 2011, Vol. 52, No. 9 Heterochromatic Flicker ERGs in Glaucoma Patients 6761

phases at 36 Hz (left) and 12 Hz (right) as a function of age forthe controls and patients, respectively. No correlation wasfound between age and response phases in either controls(P � 0.4) or patients (P � 0.2).

Although no amplitude differences were found betweensubject groups, the same analysis was performed for the re-sponse amplitudes (data not shown), and no correlation wasfound in controls (P � 0.2) or patients (P � 0.7).

DISCUSSION

Phase differences were evident in glaucoma patients at 36 Hz(for the perimetric group in the first harmonic and for bothglaucoma groups in the second harmonic) and 12 Hz (only forthe perimetric group in both the first and second harmonics).No amplitude differences were observed. In addition, the vi-sual field loss was correlated with the response phases inconditions in which perimetric and preperimetric patientsdisplayed different response phases.

The first harmonic components of responses to heterochro-matic modulation were previously proposed to reflect theactivity of the luminance mechanism at high temporal frequen-cies and the red-green cone opponent mechanism at intermediatetemporal frequencies (12 Hz).36,37 At 12 Hz, the flicker ERGthat uses L- and M-cone–isolating stimuli reflects postreceptoralmechanisms that belong to the cone opponent channel be-cause L- and M-cone–driven signals are in counterphase witheach other and have approximately equal amplitudes, indicat-ing cone opponent response properties. Less inter-individualvariability is also observed at 12 Hz, and it is less influenced bychromatic adaptation. At higher temporal frequencies (approx-imately 30 Hz and higher), the flicker ERG for L- and M-cone–isolating stimuli may be mediated by the luminance mechanismbecause the modulation is beyond the fusion frequency ofchromatic flicker. The physiological implication is that theflicker ERG that uses L- and M-cone–isolating stimuli is assumedto be determined by M pathway activity at higher temporalfrequencies and may reflect P pathway activity at 12

FIGURE 6. First harmonic (left) andsecond harmonic (right) responsephases at 12 Hz [condition R/(R�G) �0] as a function of the visual field index:mean deviation. Circles: perimetric pa-tients; triangles: preperimetric patients.The parameters used for this compari-son were significantly different betweenperimetric and preperimetric patients.There were significant correlations be-tween mean deviation values and re-sponse phases for both first (P � 0.01and r2 � 0.25) and second (P � 0.04and r2 � 0.15) harmonic responsephases.

FIGURE 7. Individual first (upper)and second (lower) harmonic re-sponse phases at 36 Hz (left) and 12Hz (right) in controls as a function ofage. No correlation was found be-tween age and response phases (P �0.6, Pearson correlation).

6762 Barboni et al. IOVS, August 2011, Vol. 52, No. 9

Hz.36,37,41–43 The present data show that it is possible to assessthe activity of luminance and cone opponent mechanismsusing a fast protocol that allows testing patients. In addition,we suggest that the second harmonic components may be alsoa valuable index for investigating retinal abnormalities in glau-coma.

Previous studies24,25,27,29–31,44 provided a better under-standing of the contribution of different components of flickerERG responses. However, assigning a specific response com-ponent to a group of cells was not possible. The present studyshowed that the second harmonic response amplitude wasrelatively small at 36 Hz. Nevertheless, closer inspection of thesecond harmonic components at 36 Hz showed that they andthe first harmonic component similarly depend on the stimuluscondition (Figs. 4, 5, left graphs). This indicates that the ERGresponses are not purely sinusoidal. However, intrusion fromanother pathway is not apparent. In contrast, at 12 Hz, the firstand second harmonic response components differently de-pended on the red fraction of the red-green counterphase-modulated stimulus (Figs. 4, 5, right graphs), indicating thatthey may reflect the activity of distinct pathways. At this tem-poral frequency, the second harmonic component and the36-Hz first harmonic responses similarly depended on the redfraction (Figs. 4, 5, upper left graphs). The present data, there-fore, suggest that the activity of the luminance mechanism maydetermine the second harmonic component of heterochromat-ic-modulated flicker ERGs at 12 Hz.

A previous study on L- and M-cone–driven ERGs showedthat the response phases in full-field ERGs may be a moresensitive indicator than response amplitudes of the differencesbetween control subjects and glaucoma patients (Link et al.IOVS 2007;48:ARVO E-Abstract 527). Compared with controls,glaucoma patients with visual field defects had different re-sponse phases in the ERG signals driven by the P pathway.Consistent with these findings, our data suggest that responsephases may be a more sensitive indicator of glaucomatouschanges than amplitudes. However, the study by Link et al.(IOVS 2007;48:ARVO E-Abstract 527) and the present workhave four main differences: in their study, the patients were

only perimetric glaucoma subjects, a CRT monitor was used asa stimulator, the stimuli were L- and M-cone isolating stimulirather than red-green modulations, and, finally, their stimulisilenced the rods. The present study used stimuli that in prin-ciple also stimulated rods and S-cones. By comparing the re-sponses with those in dichromats, we showed that neitherplayed a role in the responses.37 Subsequent measurementswith dichromats (unpublished data, 2010) confirmed this con-clusion. Additionally, in the present study, an LED Ganzfeldstimulator with narrow band emission spectra was used. As aresult, larger cone contrasts, higher mean luminances, andtherefore larger responses were feasible than were possiblewith a CRT monitor. The larger responses enabled a morereliable phase analysis.

The present results in glaucoma patients with visual fieldlosses (perimetric group) showed phase changes at both tem-poral frequencies, 36 and 12 Hz, which may represent lumi-nance and cone opponent mechanisms, respectively. Thechanges probably occurred on a postreceptoral level; if thechanges were in the cone-driven responses, we would haveexpected to find the same differences in the response phases at12 and 36 Hz, which was not the case.

The data presented here are consistent with previous psy-chophysical8,9 and electrophysiological13,45 data that showednonselective dysfunction of retinogeniculate visual pathwaysin glaucoma patients. The glaucoma patients without visualfield defects (preperimetric group) showed phase changes forthe second harmonic responses at 36 Hz. Spiking activity in theinner retina has been shown to contribute to the secondharmonic responses of photopic flicker ERGs,25 suggestingthat these phases may be correlated with spiking activity. Thisis possibly consistent with data that showed early changes inthe pattern ERG in glaucoma patients.46,47 The responsephases of flicker ERGs have been previously shown to beuseful for detecting retinal abnormalities.48–50 The amplitudeand phase responses may represent two distinct aspects ofneural activity.

In contrast with previous data,51–54 we did not find an ageeffect either on response amplitudes or on phases. However,

FIGURE 8. Individual first (upper)and second (lower) harmonic re-sponse phases at 36 Hz (left) and 12Hz (right) in the glaucoma group (in-cluding perimetric and preperimet-ric patients) as a function of age. Nocorrelation was found between ageand response phases (P � 0.2, Pear-son correlation).

IOVS, August 2011, Vol. 52, No. 9 Heterochromatic Flicker ERGs in Glaucoma Patients 6763

age-related effects are mainly observed with short-wavelengthstimuli, possibly through age-related changes in preretinal ab-sorption, which may primarily affect S-cone–driven ERGs.55 Inthe present study, we used stimuli in which short wavelengthsare absent. This may explain the absence of a clear age-relatedeffect. Additionally, ERG changes caused by lens senescencewould affect the response amplitude but not necessarily theresponse phase. Thus, the method described here may be lesssensitive to age-related effects.

CONCLUSION

The present data showed that ERG responses to heterochro-matic stimuli might have altered phases but no amplitudechanges in perimetric glaucoma patients. Glaucoma patientswithout visual field defects (preperimetric) show phasechanges at 36 Hz in the second harmonic component, possiblyrepresenting early dysfunction. Moreover, the correlation be-tween response phases and visual field index indicates that themethod may be important for monitoring glaucoma progres-sion and possible therapeutic effects in the future. Heterochro-matic flicker ERG responses provide a noninvasive functionalassessment of the luminance and cone opponent retinal mech-anisms without sensitivity to the effects of age. We thereforeconclude that the ERG method presented here may become avaluable diagnostic tool.

Acknowledgments

The authors thank Sylvia Ruhl, Anja Erhardt, Astrid Kraus, and EdithMonczak for their collaboration with the measurements.

References

1. Morrison JC, Johnson EC, Cepurna W, Jia L. Understanding mech-anisms of pressure-induced optic nerve damage. Prog Retin EyeRes. 2005;24:217–240.

2. Quigley HA, Sanchez RM, Dunkelberger GR, L’Hernault NL, Bagin-ski TA. Chronic glaucoma selectively damages large optic-nervefibers. Invest Ophthalmol Vis Sci. 1987;28:913–920.

3. Quigley HA, Dunkelberger GR, Green WR. Chronic human glau-coma causing selectively greater loss of large optic nerve fibers.Ophthalmology. 1988;95:357–363.

4. Crawford MLJ, Harwerth RS, Smith EL 3rd, Shen F, Carter-DawsonL. Glaucoma in primates: cytochrome oxidase reactivity in parvo-and magnocellular pathways. Invest Ophthalmol Vis Sci. 2000;41:1791–1802.

5. Morgan JE, Uchida H, Caprioli J. Retinal ganglion cell death inexperimental glaucoma. Br J Ophthalmol. 2000;84:303–310.

6. Yucel YH, Zhang QA, Weinreb RN, Kaufman PL, Gupta N. Effectsof retinal ganglion cell loss on magno-, parvo-, koniocellular path-ways in the lateral geniculate nucleus and visual cortex in glau-coma. Prog Retin Eye Res. 2003;22:465–481.

7. Pokorny J, Smith VC. Psychophysical signatures associated withmagnocellular and parvocellular pathway contrast gain. J Opt SocAm A Opt Image Sci Vis. 1997;14:2477–2486.

8. McKendrick AM, Badcock DR, Morgan WH. Psychophysical mea-surements of neural adaptation abnormalities in magnocellular andparvocellular pathways in glaucoma. Invest Ophthalmol Vis Sci.2004;45:1846–1853.

9. McKendrick AM, Sampson GP, Walland MJ, Badcock DR. Contrastsensitivity changes due to glaucoma and normal aging: low-spatial-frequency losses in both magnocellular and parvocellular path-ways. Invest Ophthalmol Vis Sci. 2007;48:2115–2122.

10. Sampson GP, Badcock DR, Walland MJ, McKendrick AM. Fovealcontrast processing of increment and decrement targets is equiv-alently reduced in glaucoma. Br J Ophthalmol. 2008;92:1287–1292.

11. Wakili N, Horn FK, Junemann AG, et al. The photopic negativeresponse of the blue-on-yellow flash-electroretinogram in glauco-mas and normal subjects. Doc Ophthalmol. 2008;117:147–154.

12. Arvind H, Graham S, Leaney J, et al. Identifying preperimetricfunctional loss in glaucoma: a blue-on-yellow multifocal visualevoked potentials study. Ophthalmology. 2009;116:1134–1141.

13. Greenstein VC, Seliger S, Zemon V, Ritch R. Visual evoked poten-tial assessment of the effects of glaucoma on visual subsystems.Vision Res. 1998;38:1901–1911.

14. Castelo-Branco M, Faria P, Forjaz V, Kozak LR, Azevedo H. Simul-taneous comparison of relative damage to chromatic pathways inocular hypertension and glaucoma: correlation with clinical mea-sures. Invest Ophthalmol Vis Sci. 2004;45:499–505.

15. Heron G, Adams AJ, Husted R. Central visual fields for shortwavelength sensitive pathways in glaucoma and ocular hyperten-sion. Invest Ophthalmol Vis Sci. 1988;29:64–72.

16. Casson EJ, Johnson CA, Shapiro LR. Longitudinal comparison oftemporal-modulation perimetry with white-on-white and blue-on-yellow perimetry in ocular hypertension and early glaucoma. J OptSoc Am A Opt Image Sci Vis. 1993;10:1792–1806.

17. Papaconstantinou D, Georgalas I, Kalantzis G, et al. Acquired colorvision and visual field defects in patients with ocular hypertensionand early glaucoma. Clin Ophthalmol. 2009;3:251–257.

18. Graham SL, Drance SM, Chauhan BC, et al. Comparison of psycho-physical and electrophysiological testing in early glaucoma. InvestOphthalmol Vis Sci. 1996;37:2651–2662.

19. Yates JT, Leys MJ, Green M, et al. Parallel pathways, noise maskingand glaucoma detection: behavioral and electrophysiological mea-sures. Doc Ophthalmol. 1998;95:283–299.

20. Yu JJH, Kiyosawa M, Nemoto N, Momose K, Mori H, Mochizuki M.Correlation between frequency doubling technology perimetryand temporal frequency characteristics in early glaucoma. DocOphthalmol. 2003;107:93–99.

21. Armington JC. The Electroretinogram. New York: AcademicPress; 1974.

22. Perlman I. Relationship between the amplitudes of the b wave andthe a wave as a useful index for evaluating the electroretinogram.Br J Ophthalmol. 1983;67:443–448.

23. Viswanathan S, Frishman LJ, Robson JG, Walters JW. The photopicnegative response of the flash electroretinogram in primary openangle glaucoma. Invest Ophthalmol Vis Sci. 2001;42:514–522.

24. Bush RA, Sieving PA. Inner retinal contributions to the primatephotopic fast flicker electroretinogram. J Opt Soc Am A Opt ImageSci Vis. 1996;13:557–565.

25. Viswanathan S, Frishman LJ, Robson JG. Inner-retinal contributionsto the photopic sinusoidal flicker electroretinogram of macaques:macaque photopic sinusoidal flicker ERG. Doc Ophthalmol. 2002;105:223–242.

26. Viswanathan S, Frishman LJ, Robson JG. The uniform field andpattern ERG in macaques with experimental glaucoma: removal ofspiking activity. Invest Ophthalmol Vis Sci. 2000;41:2797–2810.

27. Kondo M, Sieving PA. Primate photopic sine-wave flicker ERG:vector modeling analysis of component origins using glutamateanalogs. Invest Ophthalmol Vis Sci. 2001;42:305–312.

28. Verma R, Pianta MJ. The contribution of human cone photorecep-tors to the photopic flicker electroretinogram. J Vis. 2009;9:9.1–9.12.

29. Pangeni G, Horn FK, Kremers J. A new interpretation of compo-nents in the ERG signals to sine wave luminance stimuli at differenttemporal frequencies and contrasts. Vis Neurosci. 2010;27:79–90.

30. Burns SA, Elsner AE, Kreitz MR. Analysis of nonlinearities in theflicker ERG. Optom Vis Sci. 1992;69:95–105.

31. Chang Y, Burns SA, Kreitz MR. Red-green flicker photometry andnonlinearities in the flicker electroretinogram. J Opt Soc Am A.1993;10:1413–1422.

32. Brainard DH, Calderone JB, Nugent AK, Jacobs GH. Flicker ERGresponses to stimuli parametrically modulated in color space.Invest Ophthalmol Vis Sci. 1999;40:2840–2847.

33. Brainard DH, Roorda A, Yamauchi Y, et al. Functional conse-quences of the relative numbers of L and M cones. J Opt Soc AmA Opt Image Sci Vis. 2000;17:607–614.

34. Kremers J, Usui T, Scholl HPN, Sharpe LT. Cone signal contribu-tions to electroretinograms in dichromats and trichromats. InvestOphthalmol Vis Sci. 1999;40:920–930.

35. Kremers J, Scholl HPN, Knau H, Berendschot TT, Usui T, SharpeLT. L/M cone ratios in human trichromats assessed by psychophys-

6764 Barboni et al. IOVS, August 2011, Vol. 52, No. 9

ics, electroretinography, and retinal densitometry. J Opt Soc Am AOpt Image Sci Vis. 2000;17:517–526.

36. Kremers J, Link B. Electroretinographic responses that may reflectactivity of parvo- and magnocellular post-receptoral visual path-ways. J Vision. 2008;8:11.1–11.14.

37. Kremers J, Rodrigues AR, Silveira LC, da Silva Filho M. Flicker ERGsrepresenting chromaticity and luminance signals. Invest Ophthal-mol Vis Sci. 2010;51:577–587.

38. Mardin CY, Horn FK, Jonas JB, Budde WM. Preperimetric glau-coma diagnosis by confocal scanning laser tomography of theoptic disc. Br J Ophthalmol. 1999;83:299–304.

39. Barboni MT, Ventura DF, Kremers J. Absence of ocular interactionin flicker ERG responses reflecting cone opponent and luminancesignals. Doc Ophthalmol. 2010;121:69–75.

40. Sharpe LT, Stockman A, Jagle H, et al. Red, green, and red-greenhybrid pigments in the human retina: correlations between de-duced protein sequences and psychophysically measured spectralsensitivities. J Neurosci. 1998;18:10053–10069.

41. Kremers J, Yeh T, Lee BB. The response of macaque ganglion cellsand human observers to heterochromatically modulated lights: theeffect of stimulus size. Vision Res. 1994;34:217–221.

42. Kremers J. The assessment of L- and M-cone specific electroreti-nographical signals in the normal and abnormal human retina.Prog Retin Eye Res. 2003;22:579–605.

43. Shady S, MacLeod DIA, Fisher HS. Adaptation from invisible flicker.Proc Natl Acad Sci U S A. 2004;101:5170–5173.

44. Kondo M, Sieving PA. Post-photoreceptoral activity dominatesprimate photopic 32-Hz ERG for sine-, square-, and pulsed stimuli.Invest Ophthalmol Vis Sci. 2002;43:2500–2507.

45. Porciatti V, Di Bartolo E, Nardi M, Fiorentini A. Responses tochromatic and luminance contrast in glaucoma: a psychophysicaland electrophysiological study. Vision Res. 1997;37:1975–1987.

46. Bach M, Sulimma F, Gerling J. Little correlation of the patternelectroretinogram (PERG) and visual field measures in early glau-coma. Doc Ophthalmol. 1997;94:253–263.

47. Bach M, Unsoeld AS, Philippin H, et al. Pattern ERG as an earlyglaucoma indicator in ocular hypertension: a long-term, prospec-tive study. Invest Ophthalmol Vis Sci. 2006;47:4881–4887.

48. Scholl HPN, Kremers J. Large phase differences between L-cone-and M-cone-driven electroretinograms in retinitis pigmentosa. In-vest Ophthalmol Vis Sci. 2000;41:3225–3233.

49. Usui T, Kremers J, Sharpe LT, Zrenner E. Flicker cone electroreti-nogram in dichromats and trichromats. Vision Res. 1998;38:3391–3396.

50. Usui T, Kremers J, Sharpe LT, Zrenner E. Response phase of theflicker electroretinogram (ERG) is influenced by cone excitationstrength. Vision Res. 1998;38:3247–3251.

51. Weleber RG. The effect of age on human cone and rod Ganzfeldelectroretinograms. Invest Ophthalmol Vis Sci. 1981;20:392–399.

52. Birch DG, Anderson JL. Standardized full-field electroretinography:normal values and their variation with age. Arch Ophthalmol.1992;110:1571–1576.

53. Trick GL, Nesher R, Cooper DG, Shields SM. The human patternERG: alteration of response properties with aging. Optom Vis Sci.1992;69:122–128.

54. Fortune B, Johnson CA. Decline of photopic multifocal electroreti-nogram responses with age is due primarily to preretinal opticalfactors. J Opt Soc Am A Opt Image Sci Vis. 2002;19:173–184.

55. Chiti Z, North RV, Mortlock KE, Drasdo N. The S-coneelectroretinogram: a comparison of techniques, normative dataand age-related variation. Ophthalmic Physiol Opt. 2003;23:370–376.

IOVS, August 2011, Vol. 52, No. 9 Heterochromatic Flicker ERGs in Glaucoma Patients 6765