Tomographic and Multifocal Electroretinographic Features of IdiopathicEpimacular Membranes

Objective  To determine the relationship between the tomographic and electrophysiologiccharacteristics of the retina with an idiopathic epimacular membrane.

Methods  Sixty patients with unilateral idiopathic epimacular membranes underwentoptical coherence tomography and multifocal electroretinography (mfERG). ThemfERGs were elicited by a pseudorandom m-sequence stimulus with 37 hexagonalelements, and the mfERGs in area 1 (central 4.1°), area 2 (ring from 4.10°-7.15°),and area 3 (ring from 7.15°-13.75°) were compared with the tomographicfeatures of the corresponding area. The data from the normal fellow eyes servedas control.

Main Outcome Measures  The retinal thickness, amplitudes, and implicit time of the mfERG.

Results  On optical coherence tomographs, the retina was thickest in area 1,followed by area 2 with low tissue reflectivity of the outer retina, and area3 was of normal thickness. Electroretinography showed the amplitude ratio(affected vs fellow eyes) of mfERGs from areas 1, 2, and 3 was significantlylower than that of the controls (P<.01), and theimplicit times were significantly delayed (P<.01).The amplitude ratio was reduced the most in area 1, and the implicit timewas delayed the most in area 3. The foveal thickness was negatively correlatedwith visual acuity (ρ = −0.46; P<.001).The mfERG amplitude in area 1 was not significantly correlated with the visualacuity.

Conclusions  It is likely that retinal thickness is correlated with neural dysfunction,but mfERGs demonstrated various physiological changes in the retina.

Idiopathic epimacular membranes develop in healthy eyes with no ocularabnormalities and are typically associated with posterior vitreous detachments.^1-3 Most patients with anidiopathic epimacular membrane are older than 50 years and are asymptomatic,but some have metamorphopsia and decreased vision.

Optical coherence tomographic (OCT) studies have shown that idiopathicepimacular membranes are not due to a wrinkling of the retinal surface butare caused by a thickening of the retina. The normal foveal depression islost, and the fovea can even protrude. The visual acuity is negatively correlatedwith foveal thickness in eyes with an idiopathic epimacular membrane.⁴

Multifocal electroretinograms (mfERGs) can be used to assess the physiologicalcondition of local retinal areas noninvasively. Since the development of themfERGs in the early 1990s,⁵ the technique hasbeen used to study various macular diseases,^6-12 glaucoma,^13-16 diabeticretinopathy,^17-20 andother diseases.^21,22 Relevantto this study, Moschos et al²³ reported onthe mfERG features of eyes with idiopathic epimacular membranes but did notcompare the characteristics of idiopathic epimacular membrane with the retinalthickness.

We have examined patients with unilateral idiopathic epimacular membranesand analyzed the relationship between the OCT-determined morphological changesand the visual acuity and the electrophysiological responses in the maculararea.

We examined 60 patients with unilateral idiopathic epimacular membranesat the Department of Ophthalmology, Gunma University Hospital, Gunma, Japan.After explaining the purpose of the study and procedures to be used, informedconsent was obtained from each patient. The procedures were conducted to conformto the tenets of the Declaration of Helsinki.

The patients (35 women and 25 men) ranged in age from 18 to 79 yearswith a mean age of 62 years. Twenty-seven (45%) of the idiopathic epimacularmembranes were in the right eyes and 33 (55%), in the left eyes. All patientshad mild distortion, blurred vision, or both and no history of other ophthalmicdiseases or relevant systemic diseases. We excluded cases with pseudoholeformation and opaque, thick epimacular membranes because white lesions inducedstray light effects on the mfERGs.²⁴ The felloweyes of all patients were normal and served as controls. The degree of senilecataract in eyes with an idiopathic epimacular membrane and in the felloweyes was similar in all patients. The largest epimacular membrane was about3 disc diameters. We measured the visual acuities of the affected and normalfellow eyes in all patients.

We prospectively examined both eyes of 60 patients with an idiopathicepimacular membrane by OCT (Humphrey model 2000; Humphrey Instruments, SanLeandro, Calif). Because of the difference in reflectivity between the vitreousand retina and between the photoreceptor layer and the retinal pigment epithelium,the retinal thickness was measured using the OCT software.

The scan length was 5.0 mm through the horizontal and vertical planes.The retinal thickness was measured at 5 points in each plane of the cross-sectionalimage using the OCT software (Figure 1A). Measurements were made at the fovea (H₁,center of area 1 of the mfERGs); 1.35 mm temporal, nasal, superior, and inferiorto the fovea (H₂, corresponding to the midpoint of area 2 of themfERGs); and at 2.3 mm temporal, nasal, superior, and inferior to the fovea(H₃, corresponding to the inner margin of area 3 of the mfERGs).We defined H₁ as the retinal thickness of area 1, H₂ asthe mean thickness of the 4 quadrants and corresponding to area 2, and H₃ as the mean thickness of the 4 quadrants and corresponding to theinner margin in area 3. The thickness of the fellow eyes was measured in thesame manner (Figure 1B).

The mfERGs were recorded with the VERIS Science 4.1 (Visual evoked responseimaging system) (Mayo, Nagoya, Japan). We used a stimulus matrix that consistedof 37 hexagonal elements with a total recording time of 3 minutes 38 secondsthat was divided into 8 segments of 27.29 seconds each. The frame rate was75 Hz, and the luminance of the black-and-white frames was 3.5 candelas (cd)per m² and 200 cd/m², respectively. The contrast was96.6%.

The amplifier was set at a gain of 100 K, and the bandpass filter wasset at 10 Hz to 300 Hz. A Burian-Allen bipolar contact lens electrode wasplaced on the test eyes after the pupils were fully dilated. The distancefrom the test eye to the stimulus monitor (32 cm × 24 cm) was 33 cm.

The trace array of mfERGs represents the waveforms extracted from the37 focal electroretinograms (ERGs) and are displayed topographically. Theresponses were grouped into 4 rings of approximately equal eccentricities,and the components of the central 3 areas were studied. Their angular sizeswere area 1, 0° to 4.10°; area 2, 4.10° to 7.15°; and area3, 7.15° to 13.75°, and the diameter of each zone was 1.22 mm, 1.51mm, and 1.96 mm, respectively (Figure 2A and B). The 5-mm OCT scan covered areas 1 and 2 and thebeginning of area 3.

The first-order kernels were extracted from the mfERG in areas 1, 2,and 3. The amplitude of the first positive peak was measured from the firstnegative trough to the peak of the first positive wave. We defined the implicittime as the time from signal onset to the first positive peak. The mfERGswere recorded with a sampling interval of 0.83 milliseconds (Figure 2C).

One iteration of the system's artifact-removal algorithm was used, whicheffectively eliminated artifacts resulting from blinks and small eye movements.We compared the amplitudes and implicit times of the mfERGs in areas 1, 2,and 3 in the eyes with idiopathic epimacular membranes with the fellow eyes.

The data from the affected eyes were statistically compared with thoseof the fellow eyes with the paired t test. P<.05 was considered to be statistically significant.

Of the 60 patients, 45 (75%) had a complete posterior vitreous detachmentand 15 (25%) had partial or no posterior vitreous detachment. A wrinklingof the internal limiting membrane was observed in 48 eyes (92%) by biomicroscopy.The best-corrected visual acuity in the eyes with an idiopathic epimacularmembrane ranged from 20/200 to 20/20, while the visual acuity in the felloweyes was 20/20 or better.

The retina was thickest in area 1 in all affected eyes with the thicknessat the fovea ranging from 230 to 740 µm (mean ± SD, 457 ±121 µm). This thickness was significantly greater than the mean ±SD thickness in the fovea of the control eyes (140 ± 19 µm; P<.001). In the 60 eyes with an idiopathic epimacularmembrane, there was a negative correlation between the fovea thickness inarea 1 and best-corrected visual acuity (Spearman rank correlation, ρ= −0.46; P <.001) (Figure 3A).

The mean ± SD retinal thickness at the midpoint of area 2 (347± 59 µm) was thinner than at the fovea, but it was thicker thanthat in the fellow control eyes (249 ± 13 µm; P<.001). At the beginning of area 3 (2.3 mm from the fovea), theretinal thickness was more normal (mean ± SD thickness, 246 ±30 µm) and did not differ significantly from the control eyes (239 ±10 µm; t = 1.93; P =.058) (Figure 4).

These data demonstrated that the thickness of the retina in area 1 wassignificantly thickened, area 2 was mixed with thickened and normal retina,and area 3 was normal.

Optical coherence tomography showed that the normal foveal depressionwas not present, and the neurosensory retina in the foveal area of eyes withan idiopathic epimacular membrane had a convex shape. The tissue reflectivityof the inner retina was normal in all 3 areas. However, the outer retina wassubstantially swollen with low tissue reflectivity in areas 1 and 2. The swellingof the outer retina was greatest in area 1, which contributed to the convexappearance of the fovea. The tissue reflectivity in area 3 was normal correspondingto the normal thickness.

We extracted the first-order kernel of the mfERGs from the affectedeyes and the fellow eyes. The means ± SDs of the amplitudes and theimplicit times in eyes with an idiopathic epimacular membrane and in the felloweyes are listed in Table 1.

There was a significant reduction in the amplitudes (P<.001) of eyes with idiopathic epimacular membrane compared withthose of the control fellow eyes in rings 1, 2, and 3. The reduction of theamplitudes in area 1 was the greatest of the 3 areas (Figure 5) with a mean ratio of 77.4% in area 1, 81.2% in area 2,and 81.3% in area 3.

The implicit times were prolonged compared with those of the felloweyes (P <.001) for all areas. The implicit timeswere prolonged by 1.55 milliseconds in area 1, 1.49 milliseconds in area 2,and 1.81 milliseconds in area 3. The difference between areas 1 and 2 wasnot significantly different, but there was a significant difference betweenareas 3 and 2 (P = .02) (Figure 6).

The correlation between the amplitude of the mfERGs in area 1 and thebest-corrected visual acuity was not significant. The amplitudes of the mfERGswere also not significantly correlated with the increased fovea thicknessin area 1 (Spearman rank correlation) (Figure3B and C).

A 64-year-old man complained of blurred and distorted vision of 10 months'duration in his right eye. He noted a recent worsening of his symptoms, andhis visual acuity was 20/40 OD and 20/20 OS. Fundus examination by biomicroscopyrevealed a transparent epimacular membrane about 1.5 disc diameters. Opticalcoherence tomography demonstrated a thickening of the retina with the outerretina having low reflectivity in the macular area (Figure 1A). The thickness was 480 µm at the fovea (H₁), 376 µm at H₂ (temporal = 350 µm, nasal =370 µm, inferior = 395 µm, and superior = 390 µm), and 301µm at H₃ (temporal = 280 µm, nasal = 270 µm,inferior = 330 µm, and superior = 325 µm). Compared with the normalleft eye (Figure 1B), the overallretinal thickness was increased with the biggest increase in area 1.

In the normal left eye, H₁ was 141 µm, H₂ was268 µm (average of 270 µm, 275 µm, 265 µm, and 260µm), and H₃ was 245 µm (average of 250 µm, 240µm, 265 µm, and 260 µm).

The mfERGs were smaller than those of the normal left eye with a ratioof 75.1% (34.8/46.3 nV/deg²) in area 1, 79.5% (23.6/29.7 nV/deg²) in area 2, and 85.3% (17.4/20.4 nV/deg²) in area 3. Theimplicit times were prolonged in each area (Figure 7).

Our results showed that in the 60 eyes with an idiopathic epimacularmembrane, the retina was thickest at the fovea (457 ± 121 µm)and was thinner at area 2 but the thickness was still thicker than that inthe normal fellow eyes. At the beginning of area 3, the mean ± SD retinalthickness (246 ± 30 µm) was not significantly different fromthe fellow eyes. In areas 1 and 2, the outer retina was swollen with low reflectivity,but the inner retina maintained normal reflectivity and normal thickness.

The visual acuity was negatively correlated with the retinal thicknessat the fovea (area 1), which is consistent with a previous report.⁴ Area 1 was 1.2 mm in diameter and included the fovealpit, which was 0.5 mm in diameter with only photoreceptors, and the fovealslope where bipolar cells and the ganglion cells are present. The swellingof the outer retinal layer with low reflectivity on OCT images indicates thatthe photoreceptors were edematous, which was most likely the cause of thevisual disturbance in these eyes.

The amplitudes of the mfERGs were reduced to 77.4% in area 1, 81.2%in area 2, and 81.3% in area 3 compared with the control fellow eyes. Theimplicit times were delayed by a mean ± SD 1.55 ± 1.71 millisecondsin area 1.

These abnormalities in the mfERG in area 1 were not correlated withthe visual acuity (Figure 3B) orwith the retinal thickness (Figure 3C).In addition, abnormalities in the mfERGs in area 3 were found but the retinalthickness was not altered (Figure 4, Figure 5, and Figure 6).

Hood et al²⁵ and Horiguchi et al²⁶ investigated the retinal origins of the differentcomponents of the mfERGs in animals using various pharmacological agents thathad selective blocking activity of specific cells. From their findings andfrom data on the origin of conventional ERGs,²⁷ theysuggested that the light-evoked increase of extracellular potassium causedby the activity of on-bipolar and off-bipolar cells leads to an influx ofpotassium into Mueller cells. The mfERGs are then generated by a potassiumcurrent sink in Mueller cells and by current source at the inner limitingmembrane (ILM), the basement membrane of Mueller cells. Histological examinationof idiopathic epimacular membranes has shown that the ILM is firmly attachedto the idiopathic epimacular membrane with numerous attachment plaques.²⁸ A recent study²⁹ reportedthat removal of the ILM prolonged the implicit times of focal macular ERGs.These observations may explain the discrepancies between the mfERGs and retinalthickness in area 3 (ie, normal retinal thickness and abnormal mfERG results).We suggest that idiopathic epimacular membranes damage the ILM, and the alterationof its normal function as a current source can lead to abnormal mfERG results.

Another possibility for the discrepancy between the abnormal mfERG resultsand normal retinal thickness in area 3 is that the idiopathic epimacular membraneaffected the inner retina and induced abnormal mfERG results. Thus, Tanikawaet al³⁰ reported abnormal oscillatory potentialsin eyes with an idiopathic epimacular membrane. Hood et al²⁵ suggestedthat activity in the inner retina contributed to the first-order kernel ofmfERGs.

The results of studies examining the correlation between visual acuityand the focal ERG from the central retina are contradictory in eyes with anidiopathic epimacular membrane. Tanikawa et al³⁰ studied30 patients with unilateral idiopathic epimacular membranes using focal macularERGs and reported that there was a significant correlation between the relativeb-wave amplitude (affected eye vs fellow eye) and the visual acuity. In contrast,Moschos et al²³ reported that the correlationbetween the mfERGs in area 1 and the visual acuity was not significant. Ourresults revealed that correlation between the visual acuity and amplituderatio (affected eye vs fellow eye) was not significant.

One possible explanation for the discrepancy between visual acuity andmfERG in area 1 may be the size of the stimulus used in these different studies.The diameter of the center stimulus in mfERGs was 4.1°. The foveal pitwas 0.5 mm in diameter and consisted only of photoreceptor cells, and thevisual acuity is closely related to the function of the central fovea. Ifa smaller stimulus size was used, a correlation might have been found. However,a previous study using 61 hexagonal elements failed to find a correlationbetween focal ERG in area 1 and the visual acuity,²³ butTanikawa et al³⁰ found a signficant correlation,even with a 10° focal stimulus.

If our mfERG was altered by the dysfunction induced by swollen conecells, there should have been a good correlation of mfERGs to the visual acuity.The possible cause for this discrepancy is that mfERG changes in area 1 werecaused not only by cone cell dysfunction but also by other changes of theretina, such as the ILM changes or inner retinal damages.

In conclusion, retinal edema in eyes with an idiopathic epimacular membraneis mainly in the photoreceptor layer. The visual acuity was significantlycorrelated with the swelling of the photoreceptors in the fovea. On the otherhand, the reduction of the mfERGs demonstrated physiological changes in theretina, possibly including cone cell dysfunction and ILM changes or innerretinal dysfunction.

Correspondence: Shoji Kishi, MD, Department of Ophthalmology, GunmaUniversity School of Medicine, 3 Showamachi, Maebashi, Gunma 371-8511, Japan(kishi@med.gunma-u.ac.jp).

Submitted for publication March 5, 2003; final revision received November6, 2003; accepted April 21, 2004.