Foveal Outer Retinal Function in Eyes With Unexplained Visual Symptoms or Acuity Loss

Objective  To determine whether foveal outer retinal dysfunction is common in eyes with unexplained visual symptoms or acuity loss.

Design  Prospective study.

Participants  Seventy-three eyes of 44 consecutive patients with unexplained visual symptoms or acuity loss, 39 eyes of 39 control subjects, and 12 eyes of 7 patients with known maculopathy.

Intervention  Foveal cone electroretinography (ERG) and letter recognition perimetry.

Main Outcome Measures  Foveal cone ERG data.

Results  Abnormal foveal cone ERG data were recorded in 35 (48%) of 73 eyes (23 [52] of 44 patients). Among these 35 eyes, amplitude was lower than in normal controls (P<.001) and was correlated with visual acuity and the number of letter recognition perimetry errors (P<.05 for both). The latter was higher in eyes with abnormal retinal responses than in symptomatic eyes with normal responses (P<.01). However, initial symptoms, visual acuity, and macular appearance did not differentiate between these 2 groups. Foveal cone ERG test vs retest data showed consistent results.

Conclusion  Foveal outer retinal dysfunction is a common underlying mechanism of previously unexplained visual symptoms or acuity loss. Foveal cone ERG testing should be considered early in the evaluation of eyes with this presentation.

PATIENTS WITH foveal outer retinal dysfunction may present with diminished acuities or with visual symptoms such as blurred vision, glare, photophobia, a preference for dim lighting conditions, central or paracentral scotomata, or color vision defects. However, retinal appearance may be deceivingly healthy.^1-5 In such patients, foveal cone electroretinography (ERG) is the diagnostic test of choice,^6,7 without which these patients may be left with the diagnosis of unexplained visual symptoms or acuity loss despite comprehensive and costly ophthalmic, neurologic, and radiological evaluations.

Results of previous studies demonstrate that foveal cone ERG testing provides an objective measurement of the electrical activity of the foveal outer retina,^6,8 that amplitude is correlated with visual acuity in some forms of maculopathy,^6,8-17 and that it can help differentiate visual loss related to maculopathy from that in amblyopia or optic neuropathy.¹⁸ Foveal cone ERG also has been shown to be a reliable tool in identifying macular dysfunction not only in eyes with overt maculopathy on ophthalmoscopy^15,18 but also in those with mild and nondiagnostic macular changes or with an entirely healthy macular appearance.¹⁹ Thus, foveal cone ERG facilitates the diagnosis of such conditions as idiopathic macular dysfunction or macular degeneration with normal fundus^1,2 and familial occult macular dystrophy.^3,4

However, no information is available regarding whether foveal outer retinal dysfunction is common in eyes with unexplained visual symptoms or acuity loss. Such information could help the practicing ophthalmologist formulate a cost-effective evaluation for these eyes. If foveal cone ERG testing can help disclose foveal dysfunction in a significant number of these eyes, this test should be performed early in their evaluation. Conversely, if foveal outer retinal dysfunction is rare, then the current frequent and early use of the low-yielding,²⁰ costly, and potentially invasive neuro-imaging studies should perhaps persist. In the present study, we recorded foveal cone ERG data from 73 eyes with unexplained visual symptoms or acuity loss to obtain this information.

We tested 73 symptomatic eyes of 44 consecutive patients, aged 10 to 80 years, evaluated from January 1996 to July 1997. Referrals to our service were made from local practices and several major medical centers after extensive but unrevealing evaluations for unexplained visual symptoms or acuity loss. Inclusion criteria included visual symptoms or acuity loss determined to be of unexplained nature by at least 2 ophthalmologists. Exclusion criteria were any of the following: best-corrected Snellen visual acuity of less than 20/300, inability to maintain fixation sufficient for reliable foveal cone ERG testing regardless of visual acuity, significant media opacities or small pupils preventing continuous observation of the foveal ERG test target on the fovea throughout testing, any intraocular surgery, intraocular pressure of 22 mm Hg or greater, overt retinopathy or maculopathy on ophthalmoscopy, or general medical conditions that may affect foveal responses, such as diabetes mellitus.²¹ Because in our experience isolated foveal outer retinal dysfunction may be a unilateral or a bilateral finding, and because the origin of this dysfunction in many patients may not be clear, we treated each eye as an independent event rather than assuming bilaterality in each patient. Both eyes of 29 patients and 1 eye of 15 patients were symptomatic eyes that met all inclusion and exclusion criteria and were included in our study.

Mean±SD patient age was 50.9±18.1 years. Twenty-one (48%) participants were male and 23 (52%) were female patients. We documented detailed medical and ocular histories before performing any test to prevent questioning bias by test results. The most common presenting symptoms were gradually increasing blur in 50 (68%) of 73 eyes and "glare" in 29 eyes (40%) (Table 1 and Table 2). The symptom of glare could include significant discomfort or slow recovery from any bright light or specifically from oncoming headlights during dusk and nighttime driving or an expressed preference for dim lighting conditions. Additional symptoms were paracentral scotomata, described as nonmobile dark, white, or missing spots, in 9 eyes (12%); similar scotomata centrally in 5 eyes (7%); incidental finding of abnormal acuity in 4 eyes (6%); difficulty reading in 4 eyes (6%); central visual distortion in 3 eyes (4%); and incidental finding of suspected abnormal color vision in 2 eyes (3%). Although most patients related these symptoms voluntarily, some symptoms were extracted after specific questioning. The most common previously known associated conditions were systemic hypertension in 8 (18%) of 44 patients and history of smoking in 6 patients (14%). Family history of visual loss was present in 5 patients (11%). In 4 of these cases, relatives from 1 or more generations experienced significant visual loss of unclear cause. In the fifth case, a relative had age-related macular degeneration.

The most common unrevealing tests previously performed by the referring ophthalmologists were Humphrey 30-2 full-threshold visual fields in 40 (55%) of 73 eyes, fluorescein angiography in 37 eyes (53%), and magnetic resonance imaging or computed tomography of the brain and orbit done with fine cuts (3.5 mm for magnetic resonance imaging and 3.0 mm for computed tomography) in 31 eyes (42%) (Table 1 and Table 2). Additional tests were full-field ERG in 16 eyes (22%), pattern visual evoked responses in 12 eyes (16%), and electro-oculography in 9 eyes (12%). These tests were performed by different laboratories following such protocols as those of the International Standardization Committee.

On examination, best-corrected Snellen visual acuity ranged from 20/20 to 20/300 (mean ± SD, 0.57±0.30; ≈20/36; Table 1 and Table 2). We found approximately 20/40 or better acuity in 43 eyes (59%). Refractive spherical equivalent ranged from -7.50 to +5.00 diopters (D) (mean ± SD, −0.35±2.30 D). Ishihara pseudoisochromatic plates and Farnsworth D-15 panel testing did not reveal any color confusion pattern consistent with an inherited color vision defect in any of the eyes tested. Visual field deficits in the central 10°×10° area of each eye were mapped by letter recognition perimetry (LRP) (Table 3).^9,22,23 This test consists of monocular identification of letters flashed consecutively and separately on a computer screen situated 30 cm from the patient's eye. The letters are flashed in random order for 20 milliseconds each to 1 of 100 locations arranged in a grid with 1° intervals and to the location of a fixation cross that disappears when a letter is flashed, for a total of 101 locations. The letters are scaled in size according to their distance from fixation to compensate for the normal decline in visual acuity with eccentricity from the foveola. Full near refractive correction is used. Black spots represent missed letters (Figure 1). The upper limit of normal is 4 errors.^9,22,23

All eyes had normal intraocular pressures and minimal or no media opacities. After pupillary dilation, we found an entirely normal retinal appearance in 55(75%) of 73 eyes and minimal and nondiagnostic macular changes—including mild retinal pigment epithelial mottling, granularity, or focal depigmentation or few small drusen—in 18 eyes (25%) (Table 1 and Table 2). These changes were previously determined by 2 or more ophthalmologists, including retinal specialists, to be insufficient to explain the extent of visual symptoms or acuity loss in the studied eyes. Unrecognized vitreomacular traction was ruled out by careful slitlamp biomicroscopy using appropriate optical means.

All foveal cone ERG testing (Table 3) was done by one of us (A.W.). Recordings were obtained in a dimly lit room with a hand-held, dual-beam stimulator-ophthalmoscope (Maculoscope Spectrum, Doran Instruments, Littleton, Mass), similar to instrumentation described previously.^6,8,21,24 Photocell light-response calibration, in addition to the stimulator-ophthalmoscope internal calibration, was routinely performed before testing each patient. Before testing, the patients were subjected to ambient room illumination with dilated pupils. Responses were elicited with a 4° white stimulus of 4.8–log troland retinal illuminance flickering at 42 Hz with a 50% duty-cycle, resulting in a mean retinal illuminance of 4.5 log trolands. The stimulus was positioned on the fovea and was centered within a 12° white, steady surround of 5.5 log trolands that eliminated stray-light stimulation. Responses were monitored with a bipolar contact lens electrode (GoldLens, Doran Instruments) on the topically anesthetized cornea. Using the stimulator-ophthalmoscope factory default software, signals were differentially amplified, smoothed by a narrow bandpass filter tuned to 42 Hz, and summed by a signal-averaging computer containing an artifact reject buffer that eliminated voltage deflections of greater than 5 µV because of eye or eyelid movements. Testing involved consecutive recording periods lasting between 60 and 120 seconds each and separated from each other by 5 to 10 seconds of darkness. In all patients, 3 or more recordings were performed until responses had stabilized to ensure response reproducibility and reliability (Figure 1). We considered stabilization to have occurred if the last recording showed a change in amplitude opposite in direction to the previous trend and amplitude and implicit time (ie, the time from stimulus onset to the corresponding cornea-positive response peak) that were not different from those of the previous response by more than 10% and 1 millisecond, respectively. This response stabilization allows for the foveal cone ERG light adaptation process to take full effect.²⁴ The foveal ERG measures studied were stabilized amplitude and implicit time; published lower limit of normal for amplitude is 0.18 µV, and upper limit of normal for implicit time is 38 milliseconds.^6,24

We documented medical, ocular, and family histories and performed psychophysical testing and complete eye examinations before foveal cone ERG testing to avoid potential bias imposed by knowledge of the latter testing results.

We compared foveal cone ERG testing results with those obtained from 39 eyes of 39 control subjects who were not related to the study patients but were matched for age, sex, refractive error, and ethnic (white) origin. We determined the false-positive rate among these healthy eyes. Inclusion criteria for controls included volunteers with no visual complaints and no ocular history except for refractive error. Exclusion criteria included those described above for the study patients as well as best-corrected visual acuity of less than 20/25 or family history of visual loss other than that related to trauma. Among these 39 controls, mean ±SD age was 49.3±17.5 years and mean ±SD refractive spherical equivalent was −0.22±1.65 D.

To determine the degree of recording variability and the false-negative rate, we tested 24 eyes of 16 patients twice within 1 to 3 months and compared the amplitude and implicit time data between the 2 testing sessions for each eye. Twelve of 24 eyes (7 patients) had known maculopathy and were not part of the study population, and the remaining 12 eyes (9 patients) had unexplained visual symptoms or acuity loss.

We studied differences in age, best-corrected Snellen visual acuity, refractive error, and the number of errors on LRP between eyes with abnormal retinal responses and symptomatic eyes with normal responses with Student nonpaired t test analysis. Differences in amplitude and implicit time between eyes with abnormal responses and eyes of controls were analyzed similarly. In eyes with abnormal responses, we studied relations between foveal cone ERG data and visual acuity and the number of errors on LRP with correlation analyses. We studied intertest differences and relations in foveal cone ERG data with Student paired t test analysis and with a correlation analysis. Differences in initial symptoms, sex, and associated conditions between eyes with abnormal responses and symptomatic eyes with normal responses were studied with χ² analysis. We performed statistical analyses on a personal computer (Macintosh, Apple Computer, Cupertino, Calif) with a statistical analysis software package (StatView SE+Graphics v.1.03, Abacus Concepts Inc, Berkeley, Calif).

This study followed the tenets of the Declaration of Helsinki. Before testing, all participants gave informed consent after our explanations regarding the nature, the possible consequences, and the possible complications of the tests used.

Data are given as mean ± SD.

Among eyes of controls, we measured amplitudes of 0.18 µV or greater in all 39 eyes (100%) and implicit times of 38.0 milliseconds or less in 38 (97%) of 39 eyes. Thus, we found amplitudes of less than 0.18 µV sufficient to define abnormal responses, a figure in agreement with previously published data.^6,24 Among the 73 eyes with unexplained visual symptoms or acuity loss, we measured averaged amplitudes of less than 0.18 µV in 35 eyes (48%) and implicit times of greater than 38.0 milliseconds in 7 eyes (10%). All latter 7 eyes demonstrated amplitudes of less than 0.18 µV as well. Accordingly, we recorded abnormal foveal cone ERG data in 35 (48%) of 73 eyes (23 [52]of 44 patients, Figure 1 and Figure 2).

Compared with eyes of controls, the 35 eyes with abnormal retinal responses had significantly lower mean amplitudes (0.33±0.10 vs 0.12±0.04 µV, respectively; P <.001; t₇₂ =12.24). Mean implicit time was not significantly different between these 2 groups, although the eyes with abnormal responses showed a trend toward slower responses (34.56±1.76 vs 36.28±5.25 milliseconds, respectively; P=.06). Mean age (55.0±15.3 years) and mean refractive spherical equivalent (−0.44±1.76 D) of the 35 abnormal eyes did not differ significantly from those of the 39 eyes of controls (see data in the "Control Methods" subsection).

Among the 35 eyes with abnormal retinal responses, we found amplitude to be significantly correlated with best-corrected Snellen visual acuity (P=.02, r =0.40), and the lower the amplitude the lower the visual acuity. However, amplitude data accounted for only 16% of the variability in visual acuity. Amplitude was also significantly correlated with the number of missed letters on LRP (P=.02, r=0.40), and the lower the amplitude the higher the number of errors. Again, amplitude data accounted for only 16% of the variability in LRP results.

We found all initial symptoms, all associated conditions, and the male-female ratio not to be significantly different among the 35 eyes with abnormal retinal responses and the 38 symptomatic eyes with normal responses. Mean age (46.9±20.1 years) and mean refractive spherical equivalent (−0.30±2.77 D) of the 38 symptomatic eyes with normal responses did not differ significantly from those of the 35 abnormal eyes (P=.07 and P =.29, respectively) or from those of the 39 eyes of controls (P=.30 and P=.21, respectively).

We found a significantly larger number of errors made within the central 10°×10° area on LRP among the 35 eyes with abnormal retinal responses than among the 38 symptomatic eyes with normal responses (mean, 49.7±32.3 vs 32.0±34.9, respectively; P<.05; t₇₁ =−2.24; Table 1 and Table 2). Moreover, when we compared the number of errors made within the central 4°×4° area only, an area more closely corresponding to the 4° size of the foveal cone ERG test target, the difference reached even higher significance (mean, 10.7±6.1 vs 6.1±6.3, respectively; P<.01; t₇₁ =−3.20).

Visual acuity of 20/40-2 or better was retained by 21 (60%) of 35 eyes with abnormal retinal responses and by 22 (58%) of 38 symptomatic eyes with normal retinal function. Although mean acuity was slightly lower in the former group (0.52±0.3 vs 0.60±0.3, respectively), the difference was not significant (P=.30). Similarly, the number of eyes demonstrating any macular changes on ophthalmoscopy was not significantly different between these 2 groups (7 vs 11 eyes, respectively; Table 1 and Table 2).

Among the 24 eyes tested twice within 1 to 3 months, we found no statistically significant differences in mean amplitude and mean implicit time between the 2 recording sessions. Although amplitude data from all 24 eyes were used for these comparisons, amplitudes in 3 of these eyes were too low to allow for reliable measurement of implicit time, excluding these 3 eyes from implicit time comparisons. Mean amplitudes and mean implicit times in the first and second test sessions were 0.130±0.080 vs 0.127±0.073 µV and 36.22±4.82 vs 36.01±4.95 milliseconds, respectively. We found significant correlations between the 2 sessions in amplitude (P<.001, r=0.89, Figure 3) and implicit time (P<.001, r =0.99). However, amplitudes were subnormal in the first testing session and normal in the second in 2 of 12 eyes with known maculopathy, and vice versa in 2 other eyes with maculopathy, suggesting a 17% rate of false-negative results for each session (Figure 3). In contrast, 9 eyes with unexplained visual symptoms or acuity loss and no previously known maculopathy had consistently abnormal retinal responses in both sessions, and the remaining 3 symptomatic eyes had consistently normal responses in both sessions. Thus, we could not demonstrate significant recording variability among eyes with unexplained visual symptoms or acuity loss.

The results of our study suggest that foveal outer retinal dysfunction, as measured by foveal cone ERG testing, can be found in as many as 48% of eyes (52% of patients) with unexplained visual symptoms or acuity loss. We also found that amplitudes were correlated with 2 subjective measures of visual function in eyes with abnormal foveal responses, namely, best-corrected Snellen visual acuity and LRP. These correlations suggest that the mechanism of visual symptoms and acuity loss in these eyes could be related at least in part to foveal outer retinal dysfunction.

The true prevalence of foveal outer retinal dysfunction among eyes with unexplained visual symptoms or acuity loss cannot be clearly established for several reasons. One is that our study population could have been affected by a biased referral to our service. A second reason is foveal cone ERG recording variability, sensitivity, and specificity. We evaluated variability by testing 24 eyes—12 with known maculopathy and 12 with unexplained visual symptoms or acuity loss—twice within 1 to 3 months. We found no statistically significant differences in mean amplitude or mean implicit time between the 2 testing sessions, suggesting low recording variability. However, we found a sensitivity rate of 83.3% among eyes with known maculopathy in each recording session. This rate is in agreement with published data. A previous study found a sensitivity rate of 66% in eyes with maculopathy and 20/40 or better acuity and a higher sensitivity rate of 86% to 91% in eyes with maculopathy and less than 20/40 acuity.¹¹ This acuity-dependent difference in sensitivity is supported by a study⁸ that showed normal foveal cone ERGs in some eyes with known maculopathy and visual acuity of 20/40 or better. In our study, 22 (58%) of 38 symptomatic eyes with normal retinal responses had approximately 20/40 or better acuity. Based on the acuity-dependent sensitivity rates, the corrected prevalence of foveal outer retinal dysfunction among eyes with unexplained visual symptoms or acuity loss in our study population may increase from 48% to 65%.

However, false-positive rates that would lower the true prevalence should also be considered. In our study, we found a specificity rate of 97.4% among eyes of controls. Published specificity rates are 92% to 95%.^11,12 Based on these rates, the corrected prevalence in our study population may decrease from 65% to 62%.

In conclusion, although the true prevalence of foveal outer retinal dysfunction among eyes with unexplained visual symptoms or acuity loss may not be clearly determined, our results suggest that foveal outer retinal dysfunction is certainly not rare among these eyes.

WE BELIEVE that malingering was not a significant factor affecting the results of our study. One of our exclusion criteria was inability to maintain fixation sufficient for reliable foveal cone ERG testing regardless of visual acuity. Our method of foveal cone ERG testing requires direct observation of the patient's fovea by the tester through a dual-beam stimulator-ophthalmoscope throughout testing. Consequently, the patient's fixation is controlled at all times. In addition, the instrument records foveal outer retinal electrical response data that are entirely objective and that cannot be falsified by the patient being tested. This is in contradistinction to tests that record subjective data, such as visual acuity, color vision, or visual fields (Table 3).

The relations between visual function measurements and foveal cone ERG data may not be straightforward. Some previous studies^6,8,10-17 found good correlations between visual acuity and amplitude data. However, amplitude may not fully predict acuity in some forms of maculopathy, such as juvenile macular degeneration⁸ and macular holes.¹¹ In our study, amplitude data in eyes with abnormal retinal responses were significantly correlated with visual acuity and with the number of errors on LRP but accounted for only 16.1% to 16.2% of the variance in both measures, suggesting that additional variables may be affecting results. Moreover, 3 of 35 eyes with foveal dysfunction (patient 3, OD and OS, and patient 23, OS; Table 1) demonstrated an apparent discrepancy between relatively preserved visual acuity, borderline or normal results on LRP, and normal macular appearance on one hand and complaints of increasing visual blur or glare and abnormal foveal cone ERG responses consistently during 2 separate testing sessions on the other hand.

One possible explanation for these apparent inconsistencies is that although foveal cone ERG test target size is 4° (12.6° area), the retinal central 1° to 2° (0.8°-3.1° area) are sufficient to maintain visual acuity within normal limits.²⁵ Thus, if enough cones are functioning normally within the central 4°, although cones are damaged in the central 1° to 2°, foveal cone ERG may still be within normal range, whereas acuity is reduced. Conversely, if cones in the central 1° to 2° are still normal while many of the surrounding cones are affected, acuity may be normal in the presence of abnormal retinal responses and otherwise unexplainable visual symptoms. Similarly, LRP assesses the central 10°×10° (100°) area, an area 8 times larger than that tested with the currently used foveal cone ERG stimulator-ophthalmoscope (Figure 1). Indeed, the difference in the number of missed letters on LRP between eyes with abnormal and normal responses was more significant when only results from the central 4°×4° (16°) area rather than from the full 10°×10° area were compared between the 2 groups. An alternative explanation is that dysfunctional cones respond differently to the 42-Hz flickering light of the foveal cone ERG stimulator-ophthalmoscope than to the high-contrast, nonflickering Snellen chart letters or the letters flashed on LRP. In view of all these possible explanations, the notion that foveal cone ERG amplitude data, obtained with the currently used stimulator-ophthalmoscope, should account alone for all variability in central visual function may not be justified.

Our results also showed that visual acuity did not differ significantly between eyes with abnormal foveal responses and symptomatic eyes with normal responses. This may not be surprising because the eyes with abnormal responses were compared not with asymptomatic eyes with normal visual function but rather with symptomatic eyes with subjectively abnormal vision. These latter eyes remained undiagnosed after our evaluation and could represent foveal cone ERG false-negative results, or nonretinal causes of visual loss or symptoms such as unrecognized refractive errors or media abnormalities, optic neuropathy, or functional visual loss. Regardless of the cause of visual loss, the mean initial acuity was abnormal and was not necessarily different from that of eyes with foveal outer retinal dysfunction.

In summary, while previous studies demonstrated foveal outer retinal dysfunction in some eyes with unexplained visual loss, our results suggest that this type of previously unsuspected retinal dysfunction is in fact a common underlying mechanism of unexplained visual symptoms or acuity loss. Our study also shows that neither initial symptoms or level of visual acuity nor macular appearance can differentiate between eyes with and without foveal outer retinal dysfunction. Because several forms of occult maculopathy have been described already,^2-4 ophthalmologists cannot exclude foveal outer retinal dysfunction and occult maculopathy without foveal cone ERG testing.

At which stage of the evaluation of these patients foveal cone ERG should be recorded is also an important issue. Some patients with foveal dysfunction in our study were previously diagnosed as having functional visual loss, including malingering, and received recommendations to seek psychiatric help. Early foveal cone ERG testing could have helped avoid these recommendations and the resulting frustration. Furthermore, before our recordings, many of our patients with foveal dysfunction had undergone extensive neurologic and radiological evaluations. Although the authors were not in control of these prestudy evaluations, fine-cut neuro-imaging studies were included. However, neuro-imaging studies previously have been shown to be low yielding,²⁰ are potentially invasive, are significantly more costly than foveal cone ERG recording (Table 4), and were unrevealing in our patients. Thus, when no neurologic signs or symptoms that may give priority to neuro-imaging studies are present, we propose that foveal cone ERG recording be used routinely early in the evaluation of eyes with unexplained visual symptoms or acuity loss.

Accepted for publication June 5, 1998.

Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Fla, May 14, 1997, and at the annual meeting of American Academy of Ophthalmology, San Francisco, Calif, October 26, 1997.

We thank Michael A. Sandberg, PhD, of Harvard Medical School for his advice in statistical analysis, and A. Douglas Mathews, Yolonda E. Howard, Jill E. Mounger, and Nancy K. Burton for their technical assistance.

Corresponding author: Asher Weiner, MD, Division of Ophthalmology, Saint Luke's Medical Center, 11311 Shaker Blvd, Cleveland, OH 44104 (e-mail: aweinerbo@aol.com).