Article Text


Spatial vision in visually asymptomatic subjects at high risk for multiple sclerosis
  1. L Frisén
  1. Institute of Clinical Neuroscience, Sahlgren’s Academy of Göteborg University, Sweden
  1. Correspondence to:
 Dr L Frisén, Neuro-ophthalmology, Sahlgren’s Hospital, SE-413 45 Göteborg, Sweden; 


Objective: To investigate the state of spatial vision in visually asymptomatic subjects at high risk for multiple sclerosis.

Methods: Fifteen subjects suffering a first neurological episode suggestive of multiple sclerosis in clinical presentation, immunological profile, and magnetic resonance imaging, were examined with a new, sensitive test of spatial vision, rarebit perimetry. None had symptoms or signs of optic neuropathy.

Results: Results of rarebit perimetry were significantly worse than those of 15 age matched normal controls (p=0.01); seven patients (47%) were outside normal limits. One patient only obtained abnormal results in high pass resolution perimetry.

Conclusions: Rarebit perimetry may help to close the sensitivity gap between clinical examinations and neuroimaging.

  • demyelination
  • multiple sclerosis
  • spatial vision
  • perimetry
  • MS, multiple sclerosis
  • MRI, magnetic resonance imaging
  • HRP, high pass resolution perimetry

Statistics from

A dvances in neuroimaging have highlighted an extensive involvement of the central nervous system at early stages of multiple sclerosis (MS), even at its very first manifestation. The first bout usually produces clinical signs attributable to a single lesion, most commonly in an optic nerve, in the brain stem cerebellum, or in the spinal cord. However, most of the lesions revealed in neuroimaging reside in the cerebral hemispheres, particularly in the periventricular regions, and are clinically silent. This silence points to the need for better tools for assessing neural function. Accurate assessment has become increasingly important with the advent of new treatment modalities.1

When considering what functions and pathways might be best accessible for refined assessment, vision and the visual pathways emerge as good candidates. Strong advantages include the possibility to probe function in small subsets of fibre bundles and the pathways’ wide extension throughout the hemispheres, including the posterior periventricular regions. Although perimetry might seem ideal for the task, standard perimetry actually has a limited ability to detect small degree neural damage. Clinicopathological correlations have shown that neural channel losses must amount to 25% to 50% before standard perimetry returns consistently abnormal results. Experimental studies have revealed similar sensitivity gaps.2 To meet the need for a more sensitive assessment, a new test principle, rarebit probing, has been developed and shown to have superior sensitivity in one model disorder—that is, chiasmal compression.3 Rarebit perimetry was used in this study of 15 subjects with a first, non-visual bout of focal, neurological disease. All had immunological and neuroimaging abnormalities predictive of MS.


Patients were recruited sequentially from Sahlgren’s Hospital’s emergency neurological service where they had presented with a first neurological episode suggestive of MS. Those who had symptoms and/or signs of acute optic neuritis were excluded. Table 1 summarises key clinical data. Cerebrospinal fluid was assayed for oligoclonal banding. Magnetic resonance imaging (MRI) used a MS protocol.

Table 1

Summary of patient data

Age matched normal controls were recruited among healthy blood donors and hospital staff. All subjects underwent a detailed clinical neuro-ophthalmological evaluation, including visual fields by high pass resolution perimetry (HRP). HRP is a computer graphics perimeter known to be sensitive to optic neuritis.4,5 All subjects were novices to clinical vision testing and all subjects consented to examination by rarebit perimetry, which was the last procedure in the clinical evaluation.

Rarebit perimetry has been described in detail elsewhere.3 The name derives from the use of test targets that present a minimum of information (“rare bits”) to the visual system. In short, the test depends on brief (200 ms) presentations of high contrast, minuscule (one half of normal minimum angles of resolution) light spots on an otherwise dark liquid crystal display, under the control of a personal computer. The test task is simply to fixate on a fixation mark and to respond whenever a light spot is seen. There are 30 test areas inside 30° of eccentricity. Each area occupies a 5° circle and each area is probed repeatedly, with ever new spot positions, for a total of 10 presentations per area. There is no conventional thresholding of level of function. Instead, the test is thought to probe the completeness of the neuroretinal matrix. This matrix is normally complete in the sense that there are neither gaps nor overlaps between receptive fields. Hence, a normal subject should obtain a hit rate close to 100% (allowance has to be made for the blindspot and angioscotomata). On the other hand, a subject who has suffered a loss of receptive fields is expected to obtain a lower hit rate, by missing a proportional number of target presentations. Lesions of the visual pathways should produce the same effect, by disconnecting receptive fields.

Results were expressed as the overall mean hit rate, excluding the two target areas closest to the blindspot. Statistical analysis depended on two tailed t tests, taking p<0.05 to be statistically significant.


Most subjects obtained closely similar results in both eyes. The study concentrates on the results from the first tested eye, to exclude inter-eye dependencies and any training effects.

The 15 control subjects presented overall mean (SD) hit rates of 96.6% (2.0). Missed presentations occurred mostly in the blindspot area and along major retinal vessels (angioscotomata). A lower normal limit was defined from these results as the mean−1.96×SD—that is, 92.7%. None of the normal subjects fell outside this limit. Test times averaged 5 min 4 s.

The 15 patients were examined a median two weeks (range 1–16) after debut of symptoms, at a mean age of 31 (8) years. All had oligoclonal banding in their cerebrospinal fluid. The median number of intracranial MS-like lesions in MRI was 5 (range 0–10); the median number of hemispheric lesions was 2 (0–9). The patients showed a larger range of overall mean hit rates than did the controls (fig 1). The group mean equalled 92.0 (5.9)% and was significantly different from that of the controls (p=0.01). Seven patients (47%) fell outside the individual control limits. Mean hit rate in the blindspot area was only 12%, attesting to good fixation. For the controls, the same statistic equalled 24%; the difference was not significant (p=0.12). The median number of test errors was the same (1) for both groups, as was the modal number (0). Mean reaction times were 0.63 (0.16) s for patients and 0.70 (0.16) for controls (p=0.31).

Figure 1

Results of rarebit perimetry. Horizontal bar represents lower limit of normal.

One single patient had a focal yet asymptomatic defect in rarebit perimetry, a homonymous upper quadrant depression. The other patients presented apparently randomly distributed losses, of low degrees. There was no central field preponderance, either in individual maps or when averaging across subjects. Averages across locations were tightly correlated with those of the controls (r=0.974). However, there were a somewhat larger number of locations where two or more presentations (out of 10) were missed among patients compared with controls (average 6.4 (3.8) v 2.6 (1.8); p=0.01).

Apart from the patient with a homonymous field defect, there were no statistically significant correlations between the rarebit results and the numbers or distribution of macroscopic lesions in MRI.

The results of HRP threshold perimetry were less striking. The instrument automatically provides a number of quantitative indices. An effective capacity index aims to assess spatial density of functional neural channels, relative to normals. Mean effective capacity for controls equalled 105 (17.3)% and for patients 97.7 (21.9)% (p=0.32). Again deriving normal limits from the control subjects, only one patient fell outside HRP normal limits.

The yield of other clinical vision examinations was otherwise small. All controls were found to be normal. All subjects had normal visual acuities and all but the patient with a homonymous field defect had symmetrical central and peripheral perception of colour saturation on informal testing.6 No instance of wasting of the retinal nerve fibre layer was identified on red-free ophthalmoscopy.7


For a plausible diagnosis of early stage demyelinating disease, this study relies heavily on recent advances in neuroimaging technology. Previous reports on spatial vision in visually asymptomatic MS patients have had to focus on later stages.8,9 There seems to exist only one closely comparable study. Lycke and coworkers10 used HRP to illuminate spatial vision in patients selected on the basis of exactly the same criteria and from the same geographical area. These authors reported a 57% prevalence of HRP results outside normal limits. However, the number of patients was small (7) and the observed prevalence might be fortuitously high. There were no statistically significant differences between previous studies and this study with respect to duration of symptoms, disability (expressed as scores on the Expanded Disability Status Scale), or mean number of MRI lesions. Hence, pooling should provide a better estimate of the true prevalence of HRP abnormality. Pooling indicated a 23% prevalence of HRP abnormality—that is, one half of the prevalence found by rarebit perimetry.

Several studies have aimed to illuminate early involvement of the visual pathways in demyelinating disease by another approach—that is, selecting subjects with unilateral optic neuritis and studying the fellow eye. Many previous studies depended on perimetric techniques that are now obsolete. Studies using modern, computer assisted perimetry have identified some abnormality in many fellow eyes.5,11–13 The Optic Neuritis Treatment Trial (ONTT) is the most representative by virtue of its size (448 subjects) and its strict quality controls. Keltner et al reported that 69% of the ONTT fellow eyes showed perimetric abnormalities.12 These were labelled minimal in 62%. One quarter (26%) of all subjects had moderate or severe fellow eye abnormality. Most of the fellow eyes were found normal at a six month follow up.14 It may be that many of these subjects in reality had had truly bilateral neuropathy. There was evidence of chiasmal or retrochiasmal involvement in 3%. Therefore, the ONTT fellow eye study cannot clearly illuminate perimetry’s capability to identify clinically silent lesions in early demyelinating disease. Furthermore, previous neurological episodes were not grounds for exclusion.

In a previous study of visual field defects from a variety of visual disorders, rarebit perimetry proved fully capable of accurate topographical diagnosis.3 However, it was unable to provide topographical information in this study, except for the single instance of a homonymous depression. The other subjects presented apparently randomly distributed losses of small areas and depths, without a central field preponderance. Hence, location(s) of injury cannot be pinpointed. This would be the expected outcome of a multiplicity of lesions along the visual pathways, where each lesion is small in relation to the pathways’ cross section. Failure to identify homonymous components among the scattered depressions observed here cannot be taken as evidence against retrochiasmal lesions because homonymity cannot be taken for granted with lesions of small spatial extents.15 Furthermore, homonymity will be difficult to recognise if disseminated lesions involve the visual pathways bilaterally. The reasoning suggests that precise correlation with macroscopic MRI results cannot be expected. Indeed, none was found, except for the patient with the homonymous field depression.

An alternative interpretation of the observed deficits is that they are non-specific corollaries of disease and attributable to a general impairment in psychophysical performance. However, the internal test evidence points in the opposite direction. The rarebit test provides three indices of general performance level—that is, numbers of procedural errors, fixation stability, and reaction times. These indices indicated that the patients tended to perform somewhat better than did the controls.

Presumably, a large majority of the subjects studied here will with time accumulate an increasing number of lesions and present the picture of clinically definite MS.16 It will be interesting to see whether this expected course is associated with deteriorating rarebit perimetry results. If the test proves capable of reflecting progression of the disease, it might provide a prototype metric of functional neural channels in MS. The low cost of the instrumentation, its simplicity of use, and the short duration of the test, would be important advantages in such a setting.

Rarebit perimetry software

The rarebit perimetry software (in Microsoft Windows format) is available free of charge from the author.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.