Color Vision Defect Type and Spatial Vision in the Optic Neuritis Treatment Trial

Transcription

1 olor Vision Defect Type and Spatial Vision in the Optic Neuritis Treatment Trial Marilyn E. Schneck*^ and Gunilla Haegerstrom-Portnoyf Purpose. To describe the types of color vision defects present in the acute phase of the disease and into recovery in the 438 participants of the Optic Neuritis Treatment Trial. Methods. Patients meeting strict eligibility criteria were seen within 8 days of the onset of symptoms and then at regular follow-up visits. t the first and 6-month visits (and subsequent annual visits), spatial vision (acuity, contrast sensitivity), visual fields, and color vision were measured. Farnsworth-Munsell 1-hue tests were scored by a variant of the method of quadrant analysis described by Smith et al (mj Ophthalmol 1985; 1: ). Results. Most persons show mixed red-green () and blue-yellow () color defects (one type predominating, accompanied by a lesser defect of the other type). defects tend to be slightly more common in the acute phase of the disease, with slightly more defects at. Persons may shift defect type over time. Defect type was not related to any of the spatial vision measures at either test time or to treatment group; however, severity of color defect was related to both spatial vision measures and treatment group. onclusions. ontrary to common clinical wisdom, optic neuritis is not characterized by selective defects. olor defect type cannot be used for differential diagnosis of optic neuritis. Invest Ophthalmol Vis Sci. 1997;38: v-lptic neuritis is an acute, inflammatory, demyelinating disease of the optic nerve that is usually accompanied by ocular pain. It affects more women than men and may be associated with multiple sclerosis. 1 lthough the most common presenting symptom is blurred vision or visual field loss, a broad range of visual functions are affected. The severity of visual loss varies from a slight visual field deficit to no light perception. The constellation of visual functions affected also varies greatly. olor vision defects are common in optic neuritis, even after resolution of the defects in spatial vision. 1 " 6 The Optic Neuritis Study 1 reported that 94% of patients had abnormal color vision in the acute phase of the disease and that about 4% had residual color defects at. 5 Despite the high frequency of color defects in optic neuritis, the nature of the defects is not well characterized. It is generally agreed that the color defect From the *Smith-Kettlewell Eye Research Institute, San Francisco, and the fschool of Oplometry, University of alifornia at erkeley, alifornia. Supported by the Smith-Kettlewell Eye Research Institute. Svbmitted for publication pril 1, 1997; revised May 27, 1997; accepted May 27, Profxrietary interest category: N. Reprint requests: Marilyn Schneck, Smith-Kettlewell Eye Research Institute, 2232 Webster Street, San Francisco, in optic neuritis resembles acquired rather than congenital color defects, but the type of acquired defect is a matter of disagreement. Kollner's rule states that blue-yellow () color defects arise in disorders of the retina, whereas conditions affecting the optic nerve result in red-green () color defects. 7 Many authors report that optic neuritis produces a type II defect. 8 "" 15 However, the finding of predominantly defects in optic neuritis is not universal. defects are often reported. 2 " 416 Nonselective () losses that is, approximately equal losses of both the and color systems have also been found using various techniques. 217 " 22 The discrepancies among these earlier reports may be due to many factors, including differences in the optic nerve diseases included (due to differences in inclusion and exclusion and diagnostic criteria), sampling biases due to small sample sizes in an inherently variable single disease entity, differences in testing times in relation to the time course of the disease, differences in the severity of the disease across studies, and differences in testing or scoring methods. The recent Optic Neuritis Treatment Trial (ONTT) 1 ' 5 longitudinally followed a large, homogeneous sample of optic neuritis patients. Strict eligibil Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 opyright ssociation for Research in Vision and Ophthalmology Downloaded From: on 2/26/218

2 olor Vision in the Optic Neuritis Treatment Trial 2279 ity and exclusion criteria were used, and each participant was tested at fixed times from the time of the initial acute attack. The Farnsworth-Munsell (FM) 1-hue test was included in a large battery of tests for assessing vision. This data set thus provides an opportunity to explore the sources of variability among earlier studies. In this article, we describe the results of analyses of the FM 1-hue tests performed at the initial visit during the acute attack of optic neuritis and at the 6- month visit, at which time spatial vision had recovered in most patients. The relation between the color vision test results and spatial vision measures (visual acuity, contrast sensitivity), and visual fields is described. The results indicate that optic neuritis does not produce a single type of color defect even in this relatively homogeneous group. We find that most persons have a mixed color defect, with a predominantly or defect and a smaller but substantial defect of the other type, both during the acute attack and during recovery. t the time of the acute attack, a slight majority of selective color defects were ; at, a slight majority were. Some persons changed defect type over time. The discrepancies among earlier studies thus most likely reflect true interindividual variation in the nature of the color defect, rather than differences in methodology or spatial vision impairment at the time of testing. METHODS Patients The ONTT 1 used strict criteria for eligibility: age 18 to 45 years, visual symptoms of 8 days' duration or less, an acute clinical syndrome consistent with unilateral optic neuritis as determined by an examining neuroophthalmologist, relative afferent pupillary defect, and some definable visual field loss. detailed history of visual symptoms, pain, and past ocular, neurologic, and systemic disease was obtained, and ocular and vision function examinations were performed. Initially, 457 patients were recruited into the study. Nine were subsequently determined to be ineligible and were not enrolled into the study. Ten were lost to follow-up. The results of testing the remaining 438 patients are described here. The mean age of the patients at study entry was 31.8 (±6.7) years. Most were women (77.2%). ccording to the classification criteria of Poser et al, 23 multiple sclerosis was definite in 5.6%, probable in 7.6%, possible in 19.9%, and not present in 66.9% of subjects. The research was in compliance with the tenets of the Declaration of Helsinki. efore entry into the study, eligible patients were required to sign an informed consent form approved by the Investigational Review oard at each institution. Vision Tests Each patient received full refractive correction. Visual acuity was measured monocularly using a retroilluminated ETDRS chart 24 at 4 m. cuity was scored letter by letter and converted to a LogMR scale for analysis. Visual acuity better than 2/2 (LogMR <.) was considered normal. ontrast sensitivity was assessed using Pelli-Robson charts specially made for the ONTT 25 at 1 m (the ONTT charts have higher contrast than the now commercially available charts). Scoring was done in triplet groups of letters. Log contrast sensitivity of 1.75 or better was considered normal. Visual fields were measured with a Humphrey Field nalyzer using the 3-2 program, which assesses the central 3 of visual field. mean deviation of s 3. d was considered normal. olor vision was assessed using the Ishihara pseudoisochromatic plates and the FM 1-hue test. 26 ' 27 ll test centers used a provided standard source, a lamp with two Verilux bulbs designed to achieve a color temperature of 62 K and provide 8 lux. Room lights were off during color testing, and the testing surface was matte black. Here we restrict our description of color vision to the results obtained using the FM 1-hue test. Protocol aseline data were collected within 8 days of presentation of symptoms. Patients were then seen at days 4, 15, and 3; at weeks 7, 13, and 19; at months 6 and 12; and then yearly. t each follow-up visit, visual acuity, contrast sensitivity, and visual fields were measured. olor vision was assessed with the FM 1-hue test at baseline (during the acute attack), at, and at all subsequent annual visits. Here we consider data only from the time of the acute attack and at. nalyses of the FM 1-Hue Results Each FM 1-hue test was scored using a locally developed computer program (K. Huie, personal communication). Each test was scored along three dimensions that relate to the severity, type, and selectivity of the color defect present. The severity of the defect was determined by computing the square root of the total error score, which is the sum of the error scores on each individual cap using Kinnear's method. 28 The individual cap error score is an index of how far the cap is from the position it would occupy in a properly ordered arrangement. The criterion for normal (square root of total error score < 1.5) adopted by the ONTT was used. FM 1-hue tests with abnormal scores were further analyzed to determine defect type and selectivity. Various techniques have been developed to determine the type and selectivity of color defects measured with die FM 1-hue test. 26 ' 28 " 31 Farnsworth's original technique involved visual inspection of the polar diagram, Downloaded From: on 2/26/218

3 228 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 G Deutan Tritan Non-selective FIGURE 1. Polar plots of error scores from die FamsworthMunsell 1 hue test. () Farnsworth's original classification of type of color defect. The shaded regions indicate areas corresponding to congenital color defects as labeled. ll major error axes falling outside die shaded areas are considered nonselective. () lassification of acquired error type according to Smidi et al.si Errors falling widiin the darkly shaded zones are considered red-green (R-G); errors falling in die lighdy shaded zones are considered blue-yellow (-Y). The error pattern shown is classified as a selective blue-yellow defect using die lenient but not die strict criterion (see text). () The error pattern shown is classified as a selective blue-yellow defect using die strict criterion of Smidi et al.*' plotting the magnitude of the error for each of the 85 test caps (Fig. 1). The length of the ray extending from each cap location indicates the error score of the individual caps. The color defect is considered selective when several caps in the same region (or two regions at 18 from each other) have significant errors, provided other regions have few or no errors. The type of defect is defined by the orientation of the major axis of errors. Zones corresponding to color defects associated with loss or alteration of each of the cone types (protan, deutan, tritan) have been defined. In this scoring scheme, a major axis falling outside one of these narrow zones is classified as, even if the error pattern is highly polarized (Fig. 1). More recently, quantitative methods for determining the polarity and axis of error patterns have been developed.26'28"31 The method of Smith et al31 was selected for analysis of the ONTT data because it is appropriate for acquired color defects, whereas the other methods were developed for congenital color defects. Smith et al's technique is a form of quadrant analysis (Fig. I). Each of the 85 caps is classified as representing either the or color system processing. regions include caps 1 to 12, 34 to 54, and 76 to 84; regions include caps 13 to 33 and 55 to 75. The square root of the sum of errors on caps 1 to 12, 34 to 54, and 76 to 84 thus yield the partial error score. The square root of the sum of errors on caps 13 to 33 and 55 to 75 is the partial error score. Either or both defect types are considered to be present when the corresponding partial error scores exceed the age norms established by the authors. Thus, in this technique, a highly polar error pattern cannot be defined as on the basis of an orientation falling outside narrow predetermined regions, which is a problem with the Farnsworth method. selective defect is considered to be present when the difference between the and partial error scores exceeds criterion values. The criteria for selectivity established by Smith et al are based on the 95% confidence limits (two standard deviations [SD]) of their data. This criterion is strict: The accumulation of errors must greatly exceed that of errors for an defect to be considered selective, and vice versa. This criterion thus identifies relatively pure defects (i.e., defects having large accumulations of one error type and a much smaller number or lack of errors of the other type). We have adopted an alternative criterion based on.5 SD from Smith et al's normal mean. We chose this criterion based on ancillary analyses showing that this more lenient criterion classified error patterns as selective versus in a manner more consistent with the amplitude analysis of Knoblauch,29 the analyses of Vingrys and King-Smith,so and visual inspection of the polar plots. Figure I shows an example of an error pattern that would be classified as a selective error pattern by the.5 SD criterion (and by visual inspection), but not the 2. SD criterion. Figure 1 shows an error pattern in which the difference between and error accumulations is sufficient to meet the 2. SD criterion for a selective defect. Only 11.4% of the optic neuritis patients produced patterns with this Downloaded From: on 2/26/218

4 olor Vision in the Optic Neuritis Treatment Trial TLE l. Pass/Fail Rates on the FM 1-Hue Time of Test % Normal % bnormal % Untestable cute attack high degree of polarity at the acute attack or at 6 months. Of course, many more (76.6%) had selective defects by the lenient criterion, and results will be discussed with reference to this criterion. The pattern of results and conclusions drawn are independent of the criterion used. RESULTS During the acute attack of optic neuritis, the vast majority of subjects either failed the FM 1-hue test or had vision too poor to perform it; only 6.8% had error scores within normal limits (Table 1). Six months later, however, 6.6% passed the test, although a large group (39.4%) still had abnormal color vision. t the time of the acute attack, 39.2% of subjects could not perform the test. In the analyses described below, subjects who could perform the test at both the time of the acute attack and are considered separately (group ) from those who could perform the test only at the 6-month visit (group ). The results for the entire sample (group + group = ) are also presented where appropriate. This separation into groups allowed us to evaluate the change in color vision over time. olor Vision Defect Type The results of subjects who failed the FM 1-hue test were analyzed to determine selectivity and type of defect using the criterion described above. Table 2 shows the percentage of subjects with,, and defects during the acute attack and at. t the time of the acute attack, the frequencies of and defects were the same (29.6%). There was a tendency for more defects (4.8%). t, there was a slight tendency for more defects than defects among those in group ; 36.8% had defects, 42.1% had defects, and 21.1% had defects. The 2281 change in the distribution of color defect type from more to more failed to reach statistical significance (P =.7). t, the relative frequencies of,, and defects for those in groups and were similar to those in group ; all groups show slightly more than defects at. The shift in defect type is seen more clearly in Figure 2, which plots the distribution of orientations of major error axes (derived by Knoblauch's method 29 ) for those in group. There is a preponderance of axes in the zone at the time of the acute attack and a shift to a majority of axes at. hange in Defect Type Over Time s described above, there is a tendency for more defects at baseline but more defects at. There are several means by which the shift in the distribution may occur. Persons with defects may improve and have normal scores at, whereas defects remain stable; persons with defects at baseline may develop defects (by either an increase in the partial error score or by a decrease in the partial error score); or persons may actually shift their defect type (i.e., lose the errors in the zone and acquire errors in the zone). In fact, each of these transitions occurred (Table 3). Of the 98 subjects who had selective defects at the acute attack, 63% (62) improved and passed the FM 1-hue test at. In 6.1% (6), the initial defects became at, and 14.3% (14) who initially had defects developed defects. Of the 71 persons who had defects at baseline, 54.9% (39) improved and were color normal at 6 months. This is slightly but not significantly less than the improvement rate among those with defects initially. Of those with initial defects, 9.8% (7) became, and 15.5% (11) defects developed defects. It thus appears that a patient is approximately equally likely to switch from a to an defect (14.3%) as from an to a defect (15.5%). Of the 71 persons who had defects at the time of the acute attack, at 63.4% (45) improved and became normal, 9.8% (7) continued to show defects, 9.8% (7) developed selective defects, and 16.9% (12) developed selective defects. Of the 168 TLE 2. olor Defect Type at Time of cute ttack and at 6 Months Time of Test % % % Total Number cute attack 4.8 (98) 36.8 (35) 34.9 (22) 36.1 (57) 29.6 (71) 42.1 (4) 38.1 (24) 4.5 (64) 29.6 (71) 21.1 (2) 27. (17) 23.4 (37) = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ); = blue-yellow defect; = red-green defect; = nonselective defect. Downloaded From: on 2/26/218

5 2282 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 TLE 3. hanges in olor Vision etween Time of cute ttack and at 6 Months olor Vision at 6 Months Total at cute ttack olor Vision at cute ttack ' % N% % % % UT% Total at 6 months 1 (3) 63.3 (62) 54.9 (39) 63.4 (45) 54.8 (92) 16.3 (16) 15.5 (11) 9.8 (7) 13.1 (22) 14.3 (14) 19.7 (H) 16.9 (12) 14.3 (24) 6.1 (6) 9.8 (7) 9.8 (7) 1.1 (17) (13) 16.2 (71) 16.2 (71) 38.3 (168) 61.2 (268) 12.8 (56) 14.6 (64) 8.4 (37) 3 (13) 1% (438) % % % UT % (3) 22.4 N = normal; = blue-yellow defect; = red-green defect; = nonselective defect; UT = untestable. persons in group, at 54.8% (92) had a normal FM 1-hue test, 13.1% (22) had selective defects, 14.3% (24) had selective defects, 1.1% (17) had defects, and 7.7% (13) remained untestable. The type of defect at baseline does not predict whether color vision will be normal or abnormal at 6 months, nor does it predict the type of defect that will be present at. However, if color vision is initially normal, it will remain normal I FIGURE 2. Frequencies of major axes of the error patterns determined by Knoblauch's method encountered at baseline () and at () in group, subjects who could be measured at both times. xes falling within the darkly shaded zones are considered red-green; axes falling in the lightly shaded zones are considered blue-yellow. Spatial Vision ll spatial vision measures (contrast sensitivity, visual acuity, foveal threshold, and mean deviation of the visual fields) showed marked improvement between the time of the acute attack and (Table 4). Overall (group ), median contrast sensitivity improved by.55 log unit, acuity by.6 log unit, visual field (mean deviation) by more than 2 log units, and foveal threshold by almost 3 log units. Even those whose vision was good enough to perform the test at baseline (group ) showed considerable improvement in spatial vision at : acuity improved by.32 log unit (from ~ 2 / 3 2 to 2/15), contrast sensitivity by.5 log unit, mean deviation by ~1.4 log unit, and foveal threshold by 1. log unit. Not surprisingly, the spatial vision of the vast majority of those in group was too poor to measure at the time of the acute attack. showed marked improvement in spatial vision at. Nonetheless, acuity, contrast sensitivity, foveal threshold, and mean deviation of the visual field were all significantly worse at 6 months in group than in group (V: t<2.,aii<.rf) = 4.32, P <.1; S: t = 5.4, P <.1; FT: t = 5.48, P <.1; MD: t = 4.28, P <.1). Relation etween Frequency of olor Vision Defects and Spatial Vision The frequency of abnormal FM 1-hue test results was related to performance on each of the spatial vi- Downloaded From: on 2/26/218

6 olor Vision in the Optic Neuritis Treatment Trial 2283 TLE 4. Spatial Vision Medians at Time of cute ttack and at 6 Months cute ttack 6 Months Log S V (Log MR) MD(d) Fovthd(d) Log S V Log MR M D (d) Fov thd (d) Total Number < < < Log S = log contrast sensitivity; V = visual acuity in log MR (minimum angle of resolution); MD = mean deviation in decibels on the Humphrey 3-2 visual field; Fov thd = foveal threshold in decibels; group = testable on the FM 1-hue at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ); = blue-yellow defect; = red-green defect; = nonselective defect sion measures. This was true both at the time of the acute attack and at for all subgroups. In general, the frequency of abnormal test results increased with decreasing acuity (Table 5), mean deviation (Table 6), and foveal threshold (Table 7) of the visual field and contrast sensitivity (Table 8). Table 5 shows performance on the FM 1-hue test at the two measurement times for the three groups defined by acuity. t the acute attack (when most persons have abnormal color vision or are untestable), 65.8% of persons with normal acuity failed the test. pproximately 9% of persons with visual acuity better than 2/5 (but less than 2/2) failed the test at this time. This frequency is similar to that for persons with poor (worse than 2/5) acuity (88.2%). t, most (77%) of those in group had acuity better than 2/2; only 27% of these people failed the FM 1-hue test. Of those in group, 59% had better than 2/2 acuity at ; less than a third of these people failed the FM 1-hue test at 6 months (29.3%). Thus, the failure rate for those with 2/2 acuity at was the same for those with vision initially too poor to test and those whose spatial vision was relatively spared at the acute attack. Very few persons had acuity worse than 2/5 at ; all 21 of them failed the FM 1-hue test. cuity and color vision were thus fairly closely related at (visual acuity better than 2/2, only ~28% are abnormal; 2/2 to 2/5, ~58% are abnormal; worse than 2/5, 1% are abnormal). Table 6 shows the relation of mean deviation of the visual field to FM 1-hue performance. t baseline, the percentage of persons failing the FM 1- hue test was similar for those who had normal or moderately reduced visual fields. However, nearly all those with severely reduced visual field performance failed the FM 1-hue test initially. t the 6-month followup, performance on the FM 1-hue test was directly related to visual field sensitivity. In all three groups, the percentage of subjects failing the FM 1-hue test increased with the magnitude of the field loss. The close association is somewhat surprising, given that color vision is a foveal task and the field mean deviation primarily involves peripheral defects. Less surprising is the fact that the frequency of passing the FM 1-hue test also increased with foveal threshold, both at the time of the acute attack and at, for all groups (see Table 7). ontrast sensitivity appears to be even more strongly related to color vision than the other spatial measures (see Table 8). t both baseline and particularly at, most subjects with normal contrast sensitivity (>1.75) had normal FM 1-hue test results. The percentage of subjects with abnormal performance on the FM 1-hue test increased with decreasing contrast sensitivity. Everyone with log contrast sensitivity less than 1. failed the FM 1-hue test. Of course, the degree of correspondence among measures depends on the criterion level for normal defined for each measure. more direct, criterionfree method for evaluating the relation between color TLE 5. Relation of FM 1-Hue Performance to Visual cuity >2/2 bnormal FM 1-Hue % 2/2 > V^ 2/5 bnormal FM 1-Hue % <2/5 bnormal or NT FM 1-Hue % aseline 65.8 (41) 27. (27) 29.3 (99) 27.8 (36) 89.7 (146) 57.9 (57) 58.8 (51) 58.4 (18) 88.2 (251) 1 (5) 1 (16) 1 (21) V = visual acuity; normal FM 1-hue = square root of total error < 1.5; group = testable on the FM 1-hue at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ). Downloaded From: on 2/26/218

7 2284 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 TLE 6. Relation of FM 1-Hue Performance to Mean Deviation of the Visual Field >2/2 bnormal FM 1-Hue % -3. < MD^ -1. bnormal or NT FM 1-Hue % > -1 db bnormal or NT FM 1-Hue % aseline 7. (1) 27.2 (22) 24.7 (97) 26.4 (299) 63. (74) 54.7 (53) 66. (5) 6.2 (13) 98.2 (347) 73.3 (15) 95.2 (21) 86.1 (36) MD = mean deviation on the Humphrey visual field; normal FM 1-hue = square root of total error < 1.5; group = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ). vision and spatial vision variables is to determine the correlation between the FM 1-hue score and the spatial vision variables. Table 9 shows the correlation coefficients at baseline. Each of the spatial vision measures are significantly correlated with FM 1-hue performance at the time of the acute attack. t baseline, only subjects testable on spatial and color tests are included in the analyses. One would expect the strength of the correlations to be greater if those whose vision was too poor to perform the FM 1-hue test had been included. orrelation analyses show that contrast sensitivity is most strongly correlated with FM 1-hue performance (53% of variance in color score accounted for by contrast sensitivity). The strength of the correlation with FM 1-hue performance was similar for acuity and foveal threshold (44% and 42% of the variance accounted for, respectively). Mean deviation was the least well correlated with FM 1-hue performance; this is not surprising, because mean deviation is primarily a measure of peripheral vision function, whereas color vision is a foveal task. Table 1 gives the correlation coefficients for the three groups at. The correlation between all measures of spatial vision and FM 1-hue performance for those in group was lower than at the time of the acute attack. This is not surprising, given the improvement on all measures over time and resultant relative restriction of range of values. Nonetheless, contrast sensitivity continued to show the strongest correlation (35% of the variance in color score accounted for by contrast sensitivity). The total number of subjects included in the analyses at was 423, not 438: 13 from group still could not perform the FM-1 hue test at, and two additional persons were missing one or more spatial vision measures at. Relation of Type of olor Vision Defect to Spatial Vision Given the large improvement in spatial vision between baseline and 6-month tests, it is reasonable to ask whether the shift in predominant color defect type from to between the initial and 6-month examination is related to the improvement in spatial vision that occurs over this period. To address this issue, the relation between color defect type and spatial vision was determined at baseline and. finding of better spatial vision among subjects with rather than defects would be consistent with an explanation of changes in spatial vision underlying the observed change in color defect type over time from at baseline, when spatial vision is poor, to at, when spatial vision measures have generally recovered. Table 11 shows the mean of the spatial variables for each defect type. t baseline and, there was no significant difference in any measure of spatial vision among those with,, or defects. This finding is inconsistent with the improvement in spatial vision being responsible for the change from predominantly to defects. DISUSSION The results of the ONTT 5 support numerous earlier studies showing that color vision is significantly im- TLE 7. Relation of FM 1-Hue Performance to Foveal Threshold Time aseline >35 bnormal FM 1-Hue % 5. (32) 25.2 (214) 24. (14) 24.8 (318) 35 > Fov thd > 28 bnormal or NT FM 1-Hue % 85.6 (9) 71.4 (49) 69.8 (43) 7.6 (92) <28 bnormal or NT FM 1-Hue % 99.6 (316) 1 (7) 1 (21) 1 (28) Fov thd = foveal threshold; normal FM 1-hue = square root of total error < 1.5; group = testable at both acute attack and 6 months; group = only testable at (not testable at acute attack); group = all (group + group ). Total Number Downloaded From: on 2/26/218

8 olor Vision in the Optic Neuritis Treatment Trial 2285 TLE 8. Relation of FM 1-Hue Performance to ontrast Sensitivity LogS> 1.75 bnormal FM 1-Hue % 1.75 < LogS^ 1. bnormal or NT FM 1-Hue % Log S < 1. bnormal or NT FM 1-Hue % Total Number aseline 3. (1) 18.5 (146) 12.8 (47) 17.1 (193) 89. (212) 53.7 (121) 53.6 (11) 53.7 (231) 1 (216) 1 (3) 1 (11) 1 (14) Log S = log contrast sensitivity measured with the Pelli-Robson chart; normal FM 1-hue = square root of total error < 1.5; group = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ) paired in the vast majority of optic neuritis patients at the time of the acute attack and that residual color vision defects persist even when spatial vision recovers (i.e., when 2/2 visual acuity is regained). There is little consensus among previous studies regarding the type of color defect produced by optic neuritis;,, and defects have been reported. The purposes of the analyses presented here were to determine the sources of discrepancy in previous results and to establish a more reliable description of the color vision loss in optic neuritis. Most commonly used methods for determining the type of color defect use criterion zones defined on the basis of congenital color defects. Thus, patterns for which most errors are accumulated in a defined protan, deutan, or tritan zone are considered selective; patterns with major error axes falling outside any of these zone are considered, no matter how polarized the error pattern. There is little reason to expect that disorders of the optic nerve would produce error patterns like those produced by loss or alteration of a particular cone pigment (as occurs in congenital dichromacy or anomalous trichromacy). For this reason, use of these classification schemes can produce falsely patterns. Therefore, we chose to use the quadrant analysis of Smith et al, 31 in which highly polarized patterns cannot be classified as on the basis of axis orientation; there are no zones. Furthermore, we used a relatively lenient criterion for selectivity, reducing the required difference between and partial error scores required to classify a test result as selective. Nonetheless, at both the time of the acute attack and at into recovery, many defects were observed. This suggests that TLE 9. orrelation of olor Defect Severity With Spatial Vision at cute ttack Visual cuity (log MR) Log ontrast Sensitivity Foveal Threshold (d) Mean Deviation (d loss) both color mechanisms are affected in most people at both measurement times and that the discrepancies among earlier studies are not due to scoring criteria or sample bias, but rather reflect true inhomogeneity in this condition. There was a nonsignificant tendency for defects to occur more often at the time of the acute attack, whereas defects were more common at 6 months. The change in color defect type does not appear to be related to the degree of spatial vision impairment at the time of testing. Spatial vision improves markedly between the time of die acute attack and, suggesting that the shift in defect type was related to the improved spatial vision; however, at neither measurement time was there an association between any of the spatial vision measures (acuity, contrast sensitivity, foveal threshold, or mean deviation of the visual field) and color defect type. There was, however, a significant direct association between the severity of the spatial vision defect and the likelihood of an abnormal FM 1-hue test result. These findings suggest that the discrepancies among earlier studies were not due to differences in the severity of the spatial vision loss between subject groups at the time of testing, but may be related to when in the course of the disease patients were tested. The type of color defect at the time of the acute attack was not related to the likelihood of having normal versus abnormal color vision at. ll defect types at the time of the acute attack had similar probabilities of becoming normal at. However, there was a relation between the likelihood of a color defect at and treatment group. For those whose color vision was abnormal at the first visit or whose vision was so poor that color vision could not be tested, there was a significantly increased likelihood of having normal color vision at if given intravenous steroids compared to placebo and oral steroids. There was no difference in the rate of improvement for those receiving oral steroids over those given a placebo. This finding was also reported by the ONTT trial. 5 There was no difference between treatment groups in the relative frequencies of each defect type at. Downloaded From: on 2/26/218

9 2286 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 TLE 1. orrelation of olor Defect Severity With Spatial Vision at 6 Months Visual cuity (log MR) Log ontrast Sensitivity Foveal Threshold (d) Mean Deviation (d loss) (27) (315) (423) r T 2 r r 2 r r = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ). The shift (not statistically significant) in the distribution of defect types from more at the time of the acute attack to more at does not appear to reflect a generally faster recovery for those with rather than defects. This conclusion is supported by the observation that the rate at which subjects shifted from to defects over was similar to the rate of those switching from to defects. Subjects with defects at. the time of the acute attack were about equally likely to have as defects at. It has been suggested that the defect type in optic neuritis patients is related to the severity and type of spatial defect and fixation location. 32 " 34 We found no association between defect type and spatial vision. One may question whether this lack of difference between groups can be attributed to our relatively lenient criterion for selectivity, which results in mixed (but unequal) defects being classified as selective. To address this issue, we reanalyzed the data using the strict criterion used by Smith et al 31 (2 SD from normal), which requires much larger differences between and error scores for classification of a defect as selective. Not surprisingly, this resulted in a much larger proportion of tests being classified as, both at the time of the acute attack (84.6% versus 29.6% by the lenient criterion) and at (88.4% versus 21.1% by the lenient criterion; Table 12). y this strict criterion, most subjects had defects at the time of the acute attack and at. The pattern of results found using the strict criterion resembles that described using the more lenient criterion. t baseline, selective defects were nearly four times more likely to be (n = 29) than (n = 8). In contrast, at, defects (n = 9) were more common than defects (n = 2) in group, those who were testable at both times. The change in the distribution of color defect type over time from more to more is statistically significant (P =.3). However, given that the number of subjects with selective defects is small, these findings may be spurious. In fact, among the 63 people who could be tested only at (group ), defects denned by the strict criterion (n = 4) were about as likely as defects (n = 3). There were no significant differences in acuity, contrast sensitivity, mean deviation, or foveal thresholds of the visual field between the defect types defined by the strict criterion (Table 13). lthough subjects with defects appeared to have slightly worse TLE 11. Relation of olor Defect Type to Spatial Vision (Mean ± Standard Deviation) Time of Test Defect Type Log ontrast Sensitivity Visual cuity (Log MR) Foveal Threshold (d) Mean Deviation (d) cute 1.6 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±.18.3 ± ±.14.3 ±.2.2 ±.21.2 ±.15. ±.19.2 ±.23.2 ± ± 21.4 ± 19.6 ± 34.5 ± 33.1 ± 34.8 ± 33.1 ± 32.2 ± 33.5 ± 34. ± 32.8 ± 32.8 ± ± ± ± ± ± ± ± ± ± ± ± ± 5.9 = blue-yellow defect; = red-green defect; = nonselective defect; group = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ). Downloaded From: on 2/26/218

10 olor Vision in the Optic Neuritis Treatment Trial 2287 TLE 12. olor Defect Type at Time of cute ttack and at 6 Months: Strict riterion Time of Test cute attack % 12.1 (29) 2.1 (2) 6.3 (4) 3.8 (6) < Vc % 3.3 (8) 9.5 (9) 4.8 (3) 7.6 (12) (84) 88.9 (53) 88.6 (137) Total Number = blue-yellow defect; = red-green defect; = nonselective defect; group = testable at both acute attach and at ; group = only testable at (not testable at acute attack); group = all (group + group ). spatial vision than those with defects, none of these differences is statistically significant, as they are based on small groups, particularly at. Marre and Marre 32 have suggested that the pattern of color vision loss is related to the retinal area involved and tested; eccentric fixation caused by central visual deficits will result in impaired color vision, with relatively spared color vision, whereas preserved foveal fixation is associated with defects due to the greater susceptibility of the system to disease. Prior authors 5 ' 32 have reported such relations in small samples. The present study does not find strong evidence for such an association. Defect type denned by the strict criterion did not predict the likelihood of normal color vision at 6 months (Table 14). Nineteen of 29 (65.5%) subjects with highly selective defects and 3 of 8 (37.5%) subjects with selective defects at the time of the acute attack had normal color vision at. Interestingly, two subjects with initially highly selective defects converted to defects at that is, they lost their defects and acquired defects, suggesting a change in the population of nerve fibers affected over time. Menage et al 19 followed 3 optic neuritis patients for from the time of the initial acute attack. olor vision was assessed using the FM 1-hue test at each visit (at presentation, 6 weeks, and ). Most subjects with abnormal FM 1-hue test results showed losses at all visits. averaging of data obtained during the acute attack revealed an abnormality along the tritan axis; this group " defect" was absent at later visits. These results are in good agreement with those of this larger study. onclusions The results from this study provide no evidence to support Kollner's rule stating that defects are selectively associated with optic neuritis. Rather, the variation among earlier studies, with some reporting predominantly,, or color defects, reflects true variation among optic neuritis patients. In most instances, both the and color systems are affected, with the degree of the impairment of the two-color systems varying both between subjects and within one patient over time. The color defect type is not dependent on spatial vision at the time of die test, but it seems to shift somewhat from a higher proportion of defects in the acute phase of the disease to relatively more defects at. Thus, one must consider the time of testing relative to onset of symptoms when evaluating color vision findings in optic neuritis. It is ill advised to use color defect type as a diagnostic tool in optic neuritis. cknowledgments Data were provided by the Optic Neuritis Treatment Trial. The authors thank Dr. Joel Pokorny for helpful discussions. TLE 13. Relation of Defect Type to Spatial Vision (Mean ± Standard Deviation): Strict riterion Time of Test Defect Type Log ontrast Sensitivity Visual cuity (Log MR) Foveal Threshold (d) Mean Deviation (d) cute attack 1.9 ± ± ± ±. 1.3 ± ± ± ± ± ± ± ± ±.43.4 ± ±.38.1 ±.14.4 ±.27. ±.19.1 ±.4.27 ±.33.3 ±.2.5 ±.1.8 ±.28.1 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 5.2 = blue-yellow defect; = red-green defect; = nonselective defect; group = testable at both acute attack and at ; group = only testable at (not testable at acute attack); group = all (group + group ). Downloaded From: on 2/26/218

11 2288 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 TLE 14. hanges in olor Vision etween aseline and 6 Months: Strict riterion olor Vision at cute ttack N% % olor Vision at 6 Months % % UT % Total at cute ttack N% % % % UT% Totals at 6 months 1 (3) 65.5 (19) 37.5 (3) 6.6 (123) 54.8 (92) 61. (267) 1. (2) 2.4 (4) 1.4 (6) 6.9 (2) 25. (2) 2.5 (5) 1.8 (3) 2.7 (12) 27.6 (8) 37.5 (3) 35.5 (72) 33.3 (56) 31.7 (139).5 (1) 7.7 (13) 3.2 (14) N = normal; = blue-yellow defect; = red-green defect; = nonselective defect; UT = untestable. Key Words color vision, Farnsworth-Munsell 1-hue test, optic neuritis, Optic Neuritis Treatment Trial, spatial vision References 1. Optic Neuritis Study. The clinical profile of optic neuritis: Experience of the Optic Neuritis Treatment Trial. rch Ophthalmol. 1991; 19: Fleishman J, eck RW, Linares O, Klein JW. Deficits in visual function after resolution of optic neuritis. Ophthalmology. 1987; 94: Harrison, ecker WJ, Stell WK. olour vision abnormalities in multiple sclerosis. an J Neurosci. 1987; 14: Travis D, Thompson P. Spatiotemporal contrast sensitivity and colour vision in multiple sclerosis. rain. 1989; 112: eck RW, leary P, nderson MM Jr, Keltner JL, Shults WT, Kaufman DI. randomized, controlled trial of corticosteroids in the treatment of acute optic neuritis. The Optic Neuritis Study. N Engl f Med. 1992;326: Katz. The dyschromatopsia of optic neuritis: descriptive analysis of data from the optic neuritis treatment trial. Trans m Ophthalmol Soc. 1995;93: Kollner H. Die Storingen des Farbensinnes, Ihre Klinische edeutung und Ihre Diagnose. erlin: Karger; irch J, hisholm I, Kinnear P, et al. cquired color vision defects. In: Pokorny J, Smith V, Verriest G, Pinckers JLG, eds. ongenital and cquired olor Vision Defects. New York: Grune & Stratton; 1979: Francois J, Verriest G. Les dyschromatopsies acquises. 1. nn Oculist (Paris). 1957; 19: Francois J, Verriest G. Nouvelles observations de deficiencies acquises de la discrimination chromatique. nn Oculist (Paris). 1968;21: ox J. olour vision defects acquired in diseases of 11. the eye. rfphysiol Opt. 196; 17: ox J. olour vision defects acquired in diseases of the eye. rfphysiol Opt. 1961a; 18: ox J. olour vision defects acquired in diseases of the eye. rfphysiol Opt. 1961b; 18: Nikoskelainen E. Symptoms, signs and early course of optic neuritis. da Ophthalmol. 1975;53: Griffin JF, Wray SH. cquired color vision defects in retrobulbar optic neuritis. m f Ophthalmol. 1978;86: , Ohta Y. Studies on acquired anomalous colour vision. olour vision anomalies in patients with lesions of the retina, optic chiasma and post-occipital centre. olour 69, Gottingen; Musterschmidt; 197: Zisman F, hargava SK, King-Smith PE. Spectral sensitivities of acquired colour defectives analysed in terms of opponent colour theory. Mod Prob Ophthalmol. 1978; 19: Mullen KT, Plant GT. olour and luminance vision in human optic neuritis. rain. 1986;19:l Menage MJ, Papakostopoulos D, HartJD, Papakostopoulos S, Gogolitsyn. The Farnsworth-Munsell 1 hue test in the first episode of demyelinating optic neuritis. rf Ophthalmol. 1993; 77: Dain SJ, Rammohon KW, enes S, King-Smith PE. hromatic, spatial, and temporal losses of sensitivity in multiple sclerosis. Invest Ophthalmol Visual Sci. 199;31: Grigsby SS, Vingrys J, enes S, King-Smith PE. orrelation of chromatic, spatial and temporal sensitivity in optic nerve disease. Invest Ophthalmol Visual Sci. 1991;32: Russell MH, Murray IJ, Metcalfe R, Kulikowski JJ. The visual defect in multiple sclerosis and optic neuritis. rain. 1991; 114: Poser M, Paty DW, Scheinberg L, et al. New diagnostic criteria for multiple sclerosis: Guidelines for research protocols. nnneurol. 1983; 13: Ferris FL, Kassoff, resnick GH, ailey IL. New visual acuity charts for clinical research. m f Ophthalmol. 1982; 94: eck RW, Diehl L, leary P, the Optic Neuritis Study. The Pelli-Robson letter chart: Normative data for young adults. lin Vis Sci. 1993;8(2): Downloaded From: on 2/26/218

12 olor Vision in the Optic Neuritis Treatment Trial Farnsworth D. The Farnsworth-Munsell 1-hue and dichotomous tests for color vision. / Opt Soc m. 1943; 33: Farnsworth D. The Farnsworth-Munsell 1-Hue Test Manual, revised ed. altimore: Munsell olor o.; Kinnear PR. Proposals for scoring and assessing the 1-hue test. Vision Res. 197; 1: Knoblauch K. On quantifying the bipolarity and axis of the Farnsworth-Munsell 1-hue test. Invest Ophthalmol Visual Sci. 1987;28:77-7l. 3. Vingrys J, King-Smith PE. quantitative scoring technique for panel tests of color vision. Invest Ophthalmol Vis Sci. 1988;29: Smith V, Pokorny J, Pass S. olor-axis determination on the Farnsworth-Munsell 1-Hue test. m J Ophthalmol. 1985; 1: Marre M, Marre E. Different types of acquired color vision deficiencies on the basis of VM patterns in dependence upon the fixation mode of the diseased eye. Mod Prob Ophthalmol. 1978; 19: Pinckers, Marre M. asic phenomena in acquired colour deficiency. Doc Ophthalmol. 1983;55: Silverman SE, Hart W Jr, Gordon MO, Kilo. The dyschromatopsia of optic neuritis is determined in part by the foveal/perifoveal distribution of visual field damage. Invest Ophthalmol Vis Sci. 199;31: Downloaded From: on 2/26/218