Objective Signs of Photoreceptor Displacement by Binocular Correspondence Perimetry: A Study of Epiretinal Membranes

Abstract

purpose. To describe a method for quantitative mapping of metamorphopsia and abnormalities of oculocentric direction in subjects with epiretinal membranes.

methods. Binocular correspondence perimetry was performed using red and green dichoptic stimuli applied in a rectangular grid pattern. The study included 9 healthy subjects and 10 subjects with a unilateral premacular epiretinal membrane and a healthy fellow eye. Interocular visuospatial correspondence was expressed in a visuospatial deviation score and the binocular correspondence perimetry plots were displayed in proportion to fundus photographs. A reference interval was defined as the 95% CI for the average visuospatial deviation score in healthy subjects.

results. In 6 out of 10 subjects with epiretinal membranes, visuospatial alignment deviated beyond the reference interval found in healthy subjects, whereas 4 subjects were within the normal range. The deviation score increased with decreasing visual acuity, although indications of heterogeneity of the subject population were identified, suggesting that visual acuity reduction and metamorphopsia may be dissociated in some types of epiretinal membranes.

conclusions. Binocular correspondence perimetry enables quantitative mapping of metamorphopsia and stratification of subjects with epiretinal membranes with respect to normative references. Data from healthy subjects appear to describe a physiological level of tolerance for changes in oculocentric direction, which may apply also to the changes induced by retinal traction.

Epiretinal membrane formation ranges in severity from a subtle ophthalmoscopic cellophane sheen and no visual dysfunction to dense contracting membranes that induce metamorphopsia, cystoid retinal edema, and partial retinal detachment. ¹ Metamorphopsia is an important cause of complaints of dysfunctional binocular vision. ¹  

Ophthalmoscopic signs of retinal displacement include folds of the internal limiting membrane of the retina and/or the epiretinal membrane, displacement of vessels toward the fovea, abnormal tortuousity of retinal vessels, and cotton-wool spots. ² Using vascular bifurcations as reference points, marked changes in location have been shown before and after surgical removal of the epiretinal membrane. ² Such morphologic mapping of retinal displacement addresses only the inner layers of the retina. Metamorphopsia, however, requires displacement of photoreceptors, and it is not obvious that photoreceptors will be repositioned in the same direction and magnitude as the inner retina. 

The objective of the present study was to evaluate a novel technique for the mapping of changes in local sign or oculocentric direction (i.e., changes in the retinal location associated with a specific direction in visual space). This was done using a modification of our binocular correspondence perimetry technique developed for the study of the visuospatial distortion that follows from the centrifugal displacement of photoreceptors in eyes with a macular hole. ³ Here, we describe our findings in subjects with premacular epiretinal membranes. 

Methods

Subjects

Ten consecutive subjects with unilateral epiretinal membranes, idiopathic or associated with posterior vitreous detachment, clear refractive media, good cooperation, and a healthy fellow eye were examined. The study also included 9 healthy subjects with normal binocular vision. No participant had a history of prior eye disease or central nervous system disease. No subject had a refractive error greater than ±5 diopters or intraocular pressure over 21 mm Hg. All subjects had best-corrected visual acuity in their unaffected eye of 20/30 or better and no ocular disorder other than a unilateral epiretinal membrane. All healthy subjects had a best-corrected visual acuity of 20/20 or better in both eyes, no metamorphopsia, and normal ophthalmoscopy. All subjects were symptomatic and had metamorphopsia or blurred vision as their primary complaint at presentation. Optical coherence tomography was not used systematically in this study. The study followed the tenets of the Declaration of Helsinki and informed consent was obtained after full explanation of the nature and possible consequences of the study. 

Principle of Binocular Correspondence Perimetry

Binocular vision relies on normal retinal correspondence, but subjects with metamorphopsia often have complaints attributable to abnormal retinal correspondence and retinal rivalry. Normal retinal correspondence implies that retinal units in the two eyes projecting to the same location in space produce the perception of identical spatial location. Metamorphopsia arises when a localized shift in the spatial projection of an element of the retina occurs, presumably because of photoreceptor displacement. The displacement causes a change in oculocentric direction of the affected photoreceptors and hence interrupts the normal correspondence of retinal units. The phenomenon differs from strabismic double vision in that only a subsection of the visual field is affected and that different subsections of the visual fields are affected differently. To assess metamorphopsia and changes in oculocentric direction of eyes with epiretinal membranes, we have applied selective stimulation of corresponding retinal areas using dichoptic stimuli that can be seen with separate eyes (one stimulus can be seen with the right eye, the other with the left eye only). In a healthy subject with normal retinal correspondence, uniocular stimulation of corresponding retinal elements in each eye using two dichoptically visible stimuli sharing the same physical location will produce the impression of the same subjective spatial location. If one eye has suffered a change in oculocentric direction, the perception of the same subjective spatial location can be induced by two dichoptic stimuli that are physically separated in space, provided that binocular vision was normal before the onset of disease and that no remodeling of higher perceptual mechanisms has taken place. By measuring the physical distance between a series of dichoptically visible stimulus pairs producing the perception of the same subjective spatial location, we quantify the distorted perception in subjects with epiretinal membranes. 

Correspondence Perimetry Instrumentation and Procedure

Each stimulus pair consists of a reference stimulus (visible only to the healthy eye) and a probe stimulus (visible only to the diseased eye). We ascertained the oculocentric direction of 16 retinal image locations by stimulating, selectively, corresponding retinal points using dichoptic stimulus pairs (Fig. 1) . For each stimulus pair we measured the angular distance between the reference stimulus and the probe stimulus, the distance being quantified as the numerical sum of the vector components Ax and Ay. Subjects were seated at a slit-lamp table, leaning on a headrest and looking straight ahead at a flat white screen at 50 cm distance. A white incandescent background luminance was adjusted to 20 cd/m². Test spectacle frames mounted with appropriate lenses were used to correct for refraction and distance. A red filter glass was placed in front of one eye and a green filter glass in front of the other eye. The filters fully reject stimuli of the other color. A red diode laser (Pen-Pointer, 1 mW, 670 nm; Melles-Griot, Carlsbad, CA) and a green HeNe ion laser (05-SGR-810, 0.1 mW, 543.5 nm; Melles-Griot) provided the dichoptic stimuli. The lasers were placed on either side of the subject’s head on ball mounts that allowed full three-axis freedom of movement by manual adjustment. The size of both stimuli was 12 mm² and both stimuli subtended a visual angle of 0.46°. The stimulus luminances were adjusted using neutral density filters to correspond subjectively to Goldmann 4 e white, which approximately equals 318 cd/m². Since the eye is far more sensitive to the green 543.5 nm light than to the 670 nm red light, absolute photometric luminances were considered irrelevant. 

The pattern of stimulation was a rectangular nine-square grid, with a black cross at the center for fixation (Fig. 1) . The stimuli were applied successively at each of the 16 nodal points of the grid. The nodal points were at 4, 9, and 12° eccentricity. The central fixation target (cross) subtended 0.57° of visual angle and was fixated throughout the procedure. The subject’s response was plotted on a corresponding perimetry chart. 

For display of the reference grid and its perceived shape, the perimetry plots were inverted, scaled to match the fundus photograph, and superimposed on the fundus photograph. The centers of the fovea and the optic nerve head were used as reference points for the scaling of the fundus photograph. The blind spot was mapped during the initial phase of the procedure. Normal reference intervals were defined as the 95% CI for the average visuospatial deviation score obtained in healthy subjects. Interocular correspondence was expressed using the sum of absolute vector component values (|Ax| and |Ay|) for the reference and probe deviations, which is defined as the interocular visuospatial deviation. This parameter is an error or deviation score that increases with increasing metamorphopsia. The interocular spatial deviation was expressed as the sum of absolute vector components rather than the vector itself. This measure augments deviations that both have a vertical and horizontal component and diminishes deviations due to one-dimensional eye misalignment (heterophorias and heterotropias). 

Correspondence Perimetry Test Strategy

Throughout the procedure the subject was fixating the central target. The blind spots of both eyes were mapped using manual kinetic perimetry. The visual field within 12° eccentricity was then studied by applying a static dichoptic stimulus visible only to the healthy eye (the reference stimulus), at a node of the 16-grid reference square (Fig. 1) . A dichoptic kinetic stimulus visible only to the diseased eye (the probe stimulus) was then introduced in the periphery of the test screen. The subject was instructed to direct the examiner verbally to move the kinetic probe stimulus toward the static reference stimulus, until the two stimuli were seen by the subject to coincide. The examiner marked the location of the probe stimuli using needle marks that were invisible to the test subject. The 16 nodal positions were mapped successively in random order. The two sets of stimulus positions define the reference grid on the healthy eye and its perceptual projection on the diseased eye, the probe grid, which may be distorted to variable degrees. 

As the two stimuli approach each other, it becomes increasingly difficult to discern their relative positions. Since intense fixation often leads to fading of the stationary stimulus because of the Troxler effect, subjects were asked to blink to make the stimulus reappear. The test was completed in 15 to 20 minutes. 

Fundus Photography

Digital black and white fundus photographs in red-free illumination was made using a Topcon TRC-50X retina camera (Topcon Inc., Tokyo, Japan) equipped with a Kodak Megaplus model 1.4 digital back piece (Eastman Kodak Inc., Rochester, NY) and a PC-based image handling system (Ophthalmic Imaging Systems, Inc., Sacramento, CA). 

Results

Healthy subjects (Table 1)were able to place the kinetic stimulus over the static eccentric stimulus with an average error approaching the diameter of the stimulus (0.46°) for any position within the central 4° of the visual field. Accuracy declined with increasing eccentricity of the static stimulus. The average error in healthy subjects at 4, 9, and 12° eccentricity was 4, 6, and 8 mm, respectively, and the intersubject variation increasing with increasing eccentricity (Fig. 2) . 

Binocular correspondence perimetry in subjects with unilateral premacular epiretinal membranes demonstrated a larger variation in interocular visuospatial deviation than in healthy subjects (Tables 1 and 2)both for the central four points (P < 0.007, Mann–Whitney) and for the complete set of 16 stimulus points (P = 0.02, Mann–Whitney). 

In 6 out of 10 subjects with epiretinal membranes, interocular visuospatial deviation was outside the reference interval observed in healthy subjects, whereas 4 subjects performed within the normal range (Fig. 3) . This distribution was comparable for the four most central stimuli and for the complete set of 16 stimulus positions. Comparison of the magnitude and direction of changes in local sign (i.e., the pattern of metamorphopsia indicated by the perimetry plots) and fundus photographs of the affected eyes reveal various patterns of association between the two characteristics (Figs. 4 and 5) . 

To assess whether visuospatial misalignment could be the sole result of poor visual acuity, a Spearman’s rank correlation analysis was used to analyze this possible relationship. Visuospatial misalignment showed an insignificant negative correlation with visual acuity for the central four grid points (P = 0.23, ρ = −0.42) as well as for all 16 points (P = 0.23, ρ = −0.42). One subject (No. 8) interrupted the trend by having low acuity (20/70) in the absence of significant visuospatial deviation on binocular correspondence perimetry. After rejection of this outlier, a strong and significant correlation between deviation and visual acuity was evident for both the isolated four central points (P = 0.05, ρ = −0.66) and for the 16-point grid as a whole (P = 0.02, ρ = −0.74). 

Discussion

Binocular correspondence perimetry enables quantitative mapping of the distorted perception (metamorphopsia) induced by epiretinal membranes. For a grid pattern of stimulus locations, the change in oculocentric direction of corresponding retinal units was mapped and interocular visuospatial alignment was expressed as a one-dimensional deviation score. Comparison with healthy subjects demonstrated that epiretinal membranes were associated with both abnormal and normal deviation scores. Three subjects with normal deviation scores described metamorphopsia. Future studies should examine whether the deviation score in subjects with metamorphopsia relates to the degree of binocular visuospacial dysfunction. 

Theoretically, binocular correspondence perimetry may enable detection of metamorphopsia in cases where scattering of light in a dense membrane impairs visualization of the fine lines of the Amsler grid. A comparative validation of the two methods was made difficult by the subjective nature of the Amsler test. Obviously, the ultimate objective method should reliably predict the natural course and the effect of therapeutic interventions in terms that are meaningful for daily visual performance (e.g., binocular reading speed). 

Metamorphopsia is characterized by displacement of the projection in physical space of eccentric visual field elements, the displacement leading to a loss of correspondence with the fellow eye and distorted spatial perception (visuospatial misalignment). If epiretinal membranes are free of prismatic optical effects and the retina has not undergone inward or outward displacement, metamorphopsia could be caused by tangential retinal photoreceptor displacement and the deviation score obtained by binocular correspondence perimetry taken as a measure of this displacement. The displacement may include misalignment of the photoreceptor with respect to the center of the pupil. The longitudinal axis of each photoreceptor is aligned toward the center of the pupil, and the cells are most sensitive to light traveling along this axis (the Stiles-Crawford effect). Thus a change in orientation may alter photoreceptor sensitivity and blur its projection in visual space. However, such phenomena are unlikely to be detectable by our technique since we used suprathreshold stimuli only. 

Our ‘displacement’ theory relies on the assumption that no remodeling of higher perceptual mechanisms has taken place. Certainly, there is strong evidence that remodeling in the central nervous system does take place after retinal lesions in nonprimate mammals, ⁴ and studies on humans suggest that the receptive field of visual cortical cells is subject to expansion in response to both real and artificial scotomata. ^{5 6}  

It may be suggested that our findings could be a result of changes in receptive field properties after complete or partial deafferentiation. Although retinal or cortical remodeling, theoretically, could account for the changes expressed by the deviation score, we find convincing evidence that this explanation should be rejected. Thus, none of our subjects had a macular hole associated with their epiretinal membrane. Epiretinal membranes rarely or never cause absolute scotomata or structural neuronal damage that could stimulate regenerative growth of axons as previously described after lesions in the adult nonprimate mammal retina. ⁴ Likewise, both biomicroscopic examination and the potential for postoperative normalization indicate that no neuronal cell loss occurs in eyes with epiretinal fibrosis. Biomicroscopy often shows signs of retinal displacement, which may be more or less confined to the innermost layers of the retina. These signs comprise folds of the epiretinal membrane and inner retinal surface, abnormal tortuousity of retinal vessels that radiate toward the fovea, convexity toward the fovea of vessels that originally followed the concavity of the temporal vascular arcades, and displacement of retinal nerve fibers from the highly regular arching course they normally assume around the fovea. Fundus photography studies demonstrate that surgical membrane removal can result in the disappearance or reduction of signs of inner retinal traction, including relocation of vascular branching points toward their presumed positions of origin. ^{1 2} Consequently, the concepts that the distorted retina retains elasticity and that pathologic tension can be relieved by surgery must be accepted. There is also good evidence that visual acuity can be improved after removal of an opaque membrane. It is more difficult to ascertain the reduction of metamorphopsia, because concurrent changes in visual acuity alter the visibility of the Amsler grid and because of the subjective nature of the Amsler test. The potential benefit of binocular correspondence perimetry is that the method can map visuospatial distortion and, presumably, the relocation of the outer layers of the retina, whereas fundus photographic studies map only changes in the inner retina, which may be less closely related to visual function. 

Subjects with epiretinal membranes are challenged by two different visual problems: membrane opacity and membrane contraction. Thus, an opaque membrane that does not adhere to the fovea may reduce visual acuity by scattering of light alone, without inducing metamorphopsia. Optical coherence tomography has shown that some epiretinal membranes are clearly separated from the inner surface of the retina over a large part of their surface and that visual acuity is linearly correlated with membrane thickness. ⁷ Although visuospatial deviation was found to increase with decreasing visual acuity in our study, one subject differed markedly from the rest of the population by having poor visual acuity but normal-range visuospatial alignment. This heterogeneity of our patient population may be interpreted as suggesting that visual acuity reduction and metamorphopsia may be dissociated in some types of epiretinal membrane. 

Surgical peeling of macular epiretinal membranes is intended to eliminate traction on the retina and to arrest or reverse the process of photoreceptor displacement and the associated distortion of the perceived spatial image. Subjective improvement is often obtained, but it is unknown to what extent the surgical treatment succeeds in repositioning the photoreceptors to their original position, a question that may potentially be answered by investigations using binocular correspondence perimetry. 

The response to surgical removal of epiretinal membranes is neither uniformly favorable nor clearly understood, but the level of preoperative visual acuity and the duration of blurred vision are main prognostic factors predicting the visual outcome of surgery. ⁸ Metamorphopsia has not been quantified in clinical trials although such quantification may have a potential for determining the optimal management of epiretinal membranes. The issue is further complicated by the variable role of a given eye in a given subject’s combined binocular visual function. 

In conclusion, binocular correspondence perimetry enabled quantitative mapping of the distortion of the shape of the perceived image, clinically known as metamorphopsia. We expressed the magnitude of this distortion as a one-dimensional deviation score and interpreted abnormal scores as resulting from displacement of retinal photoreceptors caused by epiretinal membrane contraction. The perceived distortion cannot be ascribed to simple retinal or cortical remodeling. Data from healthy subjects appeared to describe a physiological level of tolerance for changes in oculocentric direction that may also apply to changes induced by tangential retinal traction. Our epiretinal membrane study population included both visuospatial deviation scores outside the normal range and scores that were within the normal range. Binocular correspondence perimetry may be useful for stratification of candidates for surgical removal of epiretinal membranes, when attempting to identify markers of outcome, and for extending the rationale for the clinical management of retinal disease with metamorphopsia. 

 Supported by the Danish Eye Health Society (Værn om Synet), the Danish Association of the Blind (Øjenfonden), and by a Patient-Oriented Diabetes Research Career Award from the Juvenile Diabetes Research Foundation (Grant No. 8–2002-130).

 Submitted for publication August 6, 2004; revised September 16 and November 1, 2004; accepted November 8, 2004.

 Disclosure: K. Krøyer, None; O.M. Jensen, None; M. Larsen, None

 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.

 

Figure 1.

 

Determining the oculocentric direction of 16 retinal image locations and their corresponding images in the fellow eye. Open circles represent reference stimuli shown dichoptically to the healthy eye. Black dots represent probe stimuli shown dichoptically to the fellow eye. One reference stimulus and one probe stimulus were shown pair-wise sequentially at each of 16 reference locations. The subject guided the probe stimulus to a location perceived to be identical with that of the reference stimulus (sharing a common subjective visual location). Left frame, the common subjective visual location of reference and probe stimuli positions as seen ideally by healthy subjects. Middle frame, reference and probe stimuli positions recorded in physical space for healthy Subject No. 7. Right frame, reference and probe stimuli positions in physical space for Patient No. 5. Perceptual distortion was assessed by measuring, for each of the 16 reference positions, the physical distance between the stimulus pairs. The reference stimulus locations were at 4, 9, and 12° eccentricity. The central fixation target (cross) was fixated throughout the procedure.

Table 1.

 

Characteristics of Healthy Subjects

Subjects											Visuospatial Deviation (|Ax| + |Ay|)
ID	Age (y)	Sex (M/F)	VA	Study Eye (R/L)	Symptoms	12 Degree Points Averaged (mm)	9 Degree Points Averaged (mm)	4 Degree Points Averaged (mm)	All 16 Points Averaged (mm)
S1	24	M	20/20	L	None	10	9	3	8
S2	22	M	20/20	L	None	7	5	4	5
S3	27	F	20/20	L	None	7	5	3	5
S4	59	M	20/20	L	None	10	7	5	7
S5	42	F	20/20	L	None	7	7	4	6
S6	34	F	20/20	L	None	6	5	4	5
S7	45	F	20/20	L	None	5	3	2	3
S8	34	F	20/20	L	None	9	8	9	9
S9	39	F	20/20	L	None	7	7	3	6
Mean						8	6	4	6
SD						2	2	2	3
Mean+ 2 SD						11	10	8	13

 Binocular correspondence perimetry was used to quantify changes in local sign for 16 stimuli points within the central 12 degrees of the visual field. The change in local sign is quantified by the physical separation between two dichoptic stimuli producing the sensation of identical visual directions. The separation between the two points, visuospatial alignment deviation, is expressed as a vector with components Ax and Ay. The table lists the absolute sum of the vector components for each stimulus pair averaged for each stimulus eccentricity and for the entire 16 points.

Figure 2.

 

Average deviation between stimulus pairs, as a function of eccentricity in 9 healthy subjects examined by binocular correspondence perimetry. The ability of healthy subjects to superimpose the mobile stimulus on the reference stimulus was seen to decrease with increasing eccentricity. Interocular visuospatial deviation was quantitated as the sum of the absolute vector components (|Ax| + |Ay|) averaged for each eccentricity. Vertical bars indicate the 95% CI of the distribution. Interocular visuospatial deviation was significantly higher for 9 and 12° compared to 4° (9°: * P < 0.02, 12°: ** P < 0.0007; Student’s t-test).

Table 2.

 

Characteristics of Patients with Unilateral Premacular Epiretinal Membranes

Patients													Visuospatial Deviation (|Ax| + |Ay|)
ID	Age (y)	Sex (M/F)	VA*	Eye (R/L)	Suspected Etiology	Symptoms	12 Degree Points Averaged (mm)	9 Degree Points Averaged (mm)	4 Degree Points Averaged (mm)	All 16 Points Averaged (mm)
PT1	60	M	20/20	R	Idiopathic	Metamorphopsia	14	9	8	10
PT2	57	F	20/25	R	PVD	Metamorphopsia	7	6	7	6
PT3	68	M	20/70	R	PVD	Blurred vision	24	18	21	20
PT4	78	M	20/50	R	PVD	Metamorphopsia	25	25	18	23
PT5	65	F	20/40	R	PVD	Metamorphopsia	21	16	10	16
PT6	67	F	20/32	L	PVD	Metamorphopsia	7	8	5	7
PT7	74	F	20/25	L	Idiopathic	Metamorphopsia	14	14	12	13
PT8	67	F	20/70	R	PVD	Metamorphopsia	6	5	4	5
PT9	69	M	20/30	L	Idiopathic	Blurred vision	8	7	4	6
PT10	77	F	20/200	R	PVD	Blurred vision	28	29	30	29
Mean							15	14	12	14
SD							9	8	9	8
Mean ± 2 SD							32	30	29	30

 See Table 1for explanation. PVD, posterior vitreous detachment;

*  VA, visual acuity in the patient’s diseased eye.

Figure 3.

 

Deviation between binocular correspondence perimetry stimulus pairs in 10 patients (outlined markers) with unilateral epiretinal membranes and 9 healthy subjects (filled markers), expressed as the numeric sum of the vector components (|Ax| + |Ay|) for all 16 reference locations (left) and for the 4 most central reference locations (right) in a rectangular grid (interocular visuospatial alignment score). The visuospatial deviation exceeded the normal range (95% CI) in the same 6 out of 10 subjects, both for the central 4 reference locations as well as for all of the 16 reference locations. Error bars, the 95% CI for the average visuospatial deviation in healthy subjects.

Figure 4.

 

Fundus images (grayscale, red-free illumination) and binocular correspondence perimetry plots from 10 eyes in 10 subjects with unilateral epiretinal membranes. The rectangular reference grids (thin line) are shown as projected on the retina and the distorted probe grids (thick line). The grids were inverted and scaled to fit the foveola-disc distance as determined by perimetric mapping of the physiological blind spot.

Figure 5.

 

Superimposition on fundus photographs from two eyes in two subjects with epiretinal membranes of reference grids and probe grids, as recorded using binocular correspondence perimetry. The grids were inverted and scaled to the magnification of the fundus image. In both cases, the distorted probe grids are in agreement with the apparent retinal traction toward the center of the fovea. Morphologic signs of traction are seen as folds in the epiretinal membrane radiating toward the foveola and abnormally fovea-convex and tortuous patterns of retinal vessel in the macula in both eyes.