Comparing Pupil Function with Visual Function in Patients with Leber’s Hereditary Optic Neuropathy

Abstract

purpose. To compare pupil function with visual function in patients with Leber’s hereditary optic neuropathy (LHON) and age-matched normal control subjects.

methods. Visual function was assessed by measuring the perceptual thresholds at five central locations in the visual field using automated static perimetry. Pupil function was assessed by recording the pupil responses to a standard intensity light stimulus (size equivalent to a Goldmann V target) presented at the same five locations in the visual field. The extent of the pupil afferent defect in LHON patients was quantified by establishing the relationship between stimulus intensity and the size of the pupil response in normal subjects and then interpolating the equivalent luminance deficit in LHON patients from the size of their pupil responses.

results. At all five locations tested, the pupil responses were significantly reduced in amplitude, and the perceptual thresholds were significantly raised in LHON patients compared with normal control subjects. A nonparametric analysis of perceptual and pupil responses to perithreshold stimuli showed that a stimulus that was not perceived was three times more likely to be followed by a pupil response in a LHON patient than in a normal subject (P < 0.001). A quantitative comparison showed that the visual deficits exceeded the pupil deficits by on average 7.5 dB at all tested locations.

conclusions. Although both visual and pupil function are abnormal in LHON, there appears to be relative sparing of the pupil afferent fibers.

Leber’s hereditary optic neuropathy (LHON) is a rare disease associated with mutations in mitochondrial rather than nuclear DNA. Affected patients typically experience profound bilateral visual loss (<6/60 in >95% of cases ¹ ) with poor recovery. The visual field most commonly shows a dense, large central scotoma. Despite the poor visual function in this group of patients, it has often been observed that the pupil shows a relatively normal response to light. ^{2 3 4} This clinical impression has been supported by some studies in which infrared videopupillography was used ⁵ but not by other studies. ⁶  

If the pupil afferent pathway is not affected in LHON, then such pupillovisual dissociation has intriguing implications regarding the mechanism of damage in this disease. This would, however, be a surprising result in that afferent pupil defects are invariably present in other optic neuropathies. An alternative explanation is that the pupil responds well in LHON patients because when tested with a full-field light stimulus, the afferent drive from the intact peripheral field is sufficient to generate apparently normal pupil reflexes. Jacobson et al. ⁷ recently reported a series of 10 cases of LHON in which vision had been lost in only one eye. Using the swinging flashlight test and a full-field stimulus, they found a relative afferent pupil defect (RAPD) in all cases and concluded that the size of the RAPD matched the extent of the visual field defect with no evidence of pupillovisual dissociation. 

In the present study we have taken a different approach to this question. Using a modified automated perimeter, we recorded the pupil responses to much smaller light stimuli that were presented at discreet locations within the visual field. By measuring the perceptual thresholds at the same locations, we were then able to compare pupil function with visual function directly in patients with LHON. Our results confirm that both pupil and visual function are abnormal at discrete locations within the central scotoma in this disease. However, both a nonparametric and a quantitative comparison of the visual and pupil deficits suggest that the pupil afferent fibers are relatively spared. A preliminary account of these findings has been published elsewhere. ⁸  

Methods

Subjects

Nineteen patients with LHON were included in this study. Mitochondrial DNA analysis confirmed point mutations at positions 11,778 (12 cases), 3460 (2 cases), and 14,484(5 cases). All patients were men, with a median age of 37 years (range, 21–71 years). In all cases the visual loss was chronic, with on average 12.4 years (range, 2–32 years) between visual loss and evaluation in this study. For comparison, we examined 24 control subjects with healthy eyes (M:F ratio, 12:11; median age, 28 years [32 years in the male subjects]), range, 21–51 years); all control subjects had normal visual function with corrected Snellen acuities of 6/6 or better. Pupil and visual function were assessed after a minimum of 10 minutes preadaptation to mesopic conditions, and only the preferred eye was tested in each case. All patient and control subjects gave their informed consent before participating in this study, and the research followed the tenets of the Declaration of Helsinki. 

Assessment of Visual Function

Distance acuity was measured at 6 meters with appropriate refractive correction using a Snellen chart. Visual thresholds were estimated using static automated perimetry (Octopus 1-2-3; Interzeag AG, Zurich, Switzerland). Five stimulus locations were tested: fixation, and at 17° eccentricity along the 45°/135° meridians in the superotemporal, inferotemporal, superonasal and inferonasal quadrants. The stimulus size and duration were 0.4° (Goldmann III) and 100 ms, respectively, with a background illumination of 31.4 apostilb (asb). Threshold determination was achieved using the 4-2-1 dB staircasing strategy over a stimulus intensity range of 100 to 4000 asb. Subjects were asked to fixate on a nonaccommodative red target in the center of the perimeter bowl: in cases where the patient was unable to see the fixation target, the patient was asked to look into the middle of the bowl using their intact peripheral vision to guide centration. Fixation was monitored throughout using a magnified infrared camera image: virtual cursors were manually adjusted before each test to set a fixation “window,” which, if transgressed by the pupil margins, would stop the test until fixation was restored. 

Assessment of Pupil Function

Pupil function was also evaluated with the Octopus 1-2-3 automated perimeter using the same background illumination of 31.4 asb. The pupil reflex response to a light stimulus presented at each of the above five locations was recorded using an infrared video camera (temporal resolution, 19.38 ms; spatial resolution, 0.05 mm). Throughout each test, the light stimulus was presented repeatedly, and the sequence of test locations and interstimulus intervals varied (mean interval, 5.0 seconds; range, 4.0–6.0 seconds) using computer software. The stimulus parameters for intensity, size, and duration were changed for pupil perimetry to 4000 asb, 1.7° (Goldmann V), and 500 ms, respectively, to ensure reliable pupil reactions. The blink rate was minimized by instilling a single drop of oxybuprocaine 0.4% in both eyes. Each recording trial lasted 2 minutes; the subject was then rested between trials to maintain optimum alertness. The pupil size was monitored in real time throughout each recording trial using the magnified infrared camera image incorporated into the Octopus 1-2-3 perimeter. Off-line measurements of the baseline pupil diameter preceding each stimulus presentation were obtained by computer analysis of the video images and in all cases were found to lie within the normal age-matched range. In a few cases fatigue waves were clearly visible in the recording: these data were discarded, and the recording was repeated when the subject was more alert. 

Comparing Pupil and Visual Function

We used two different techniques for comparing pupil and visual responses. First, during each pupil test the subject was instructed to press a response button if the stimulus was seen. The pupil recordings were examined independently and off-line by two of the authors (FB and JS) and judged as showing either a reflex response to the stimulus or no response. Nonparametric statistics were then used to compare the pupil and perceptual responses to perithreshold light stimuli in both LHON patients and normal control subjects. 

A second, quantitative technique used the principle of a pharmacologic assay. In normal control subjects we randomly varied the stimulus intensity over a 40-dB range (fixation) or 20-dB range (eccentric locations) to characterize (for each stimulus location) the relationship between stimulus intensity and the amplitude of the pupil response. We then used this relationship to interpolate the amount by which you would have to reduce the stimulus intensity in a normal control subject to produce a pupil response equivalent to that observed in the patient: this value corresponds to the size of the pupil deficit and may be compared directly with estimates of the visual deficit. 

Data Analysis

The pupil recordings were analyzed on a computer using commercially available curve-fitting software (Interzeag AG) to measure the amplitude of the reflex response. For each subject standard descriptive statistics were used to derive the best estimate of the pupil response amplitude when stimulating each of the five locations. Pupil response amplitudes were compared at different stimulus locations and between patients and control subjects using Student’s t-tests. The proportion of perithreshold stimuli that could be seen by the subject and the proportion of these stimuli that produced pupil responses were compared in the LHON and the normal (control) study groups using the χ² statistic. A goodness-of-fit-test was used to determine whether the differences between visual and pupil deficits were normally distributed. The mean pupil deficit was compared with the mean visual deficit at each stimulus location using the Wilcoxon signed-ranks test. 

Results

Visual Function

All LHON patients in this study had poor vision. The median corrected Snellen acuity was 1/60 (range, 6/9 to HM [hand movements]) with 74% of eyes tested seeing worse than 6/60. Threshold determinations in the LHON patients are summarized in Figure 1 , which shows the mean difference (±95% confidence intervals) between the patient threshold and the normal threshold at each of the five locations tested. The greatest deficit was seen at fixation where the visual sensitivity was reduced by on average 27 dB. However, even at 17° eccentricity we found an average difference of between 17 and 18 dB between the patient threshold and the normal threshold. 

Pupil Function

The pupil responses to standard intensity (4000 asb) stimuli presented at the same five locations were tested in both patients and normal control subjects. The mean amplitude of these responses (expressed as percentage constriction of the pupil area ± 95% confidence intervals) is shown in Figure 2 . In control subjects, the largest pupil responses were obtained when the stimulus was presented at fixation. When the same intensity of stimulus was presented at 17° eccentricity, the pupil responses were significantly smaller (mean difference, 46%, P < 0.001) with a tendency for stimuli presented in the inferonasal quadrant to elicit the smallest responses. No difference was found between male and female control subjects. In LHON patients, the pupil responses were significantly smaller (P < 0.001) than in normal control subjects at all five of the locations tested. The magnitude of this difference was greatest for stimuli presented at fixation (response amplitude on average was 65% smaller in LHON patients than in normal control subjects) and less marked at 17° eccentricity (42% smaller). 

Nonparametric Comparison of Visual and Pupil Responses

Pupil responses were observed in LHON patients after subthreshold stimuli. An example is shown in Figure 3 . A standard intensity light stimulus (4000asb) was presented 14 times to the same location in the superotemporal quadrant; pupil responses were present on every occasion with amplitudes ranging from 5% to 21% (mean, 11%), and yet the patient could not perceive any of these stimuli. In the course of this study we found seven similar cases in which the standard intensity light stimulus was never perceived by the patient: in all seven cases, pupil responses were observed with frequencies ranging from 9% to 100% (median, 53%). In normal control subjects the standard intensity stimulus always was perceived during the pupil test, but when the stimulus intensity was reduced by 40 dB (fixation) or 20 dB (eccentric locations), there were seven cases in which the light stimulus was never perceived. In contrast to LHON patients, in four of these seven cases no pupil responses were observed after subthreshold stimuli; in the remaining three cases, pupil responses were only rarely observed (frequencies in the range 6% to 15%). 

When all the perceptual responses to standard intensity light stimuli in LHON patients are considered, there were 169/759 (22%) stimuli that were not perceived, of which 89/169 (53%) were followed by a pupil response. In normal control subjects using perithreshold light intensities, a similar proportion of stimuli were not perceived (133/645, 21%), but the proportion of these subthreshold stimuli that were followed by a pupil response was three times lower (20/133, 15%). A χ² test confirms that the difference in results between the two study groups was highly significant (χ² = 32.17, P < 0.001). 

A Quantitative Comparison of the Visual Deficit and the Pupil Deficit

By varying the stimulus intensity, we characterized the relationship between stimulus intensity and response size in normal control subjects at each of the five tested locations. The results for stimuli presented at fixation are illustrated in Figure 4 . The relationship is sigmoidal, with the lowest intensity stimuli producing pupil responses more than 10 times smaller than those following our “standard” intensity stimulus (4000 asb, which corresponds to the 0-dB attenuation in Fig. 4 ). Similar relationships were found at the four eccentric locations. The pupil deficits in LHON patients were interpolated from these graphs and have been plotted against the corresponding estimates of visual deficit in Figure 5 . 

Most of the data points lie below the line of unit gradient (i.e., in most cases the visual deficit exceeded the pupil deficit). We could find no evidence to suggest that this general result applied only to certain locations in the visual field, or to particular mitochondrial mutations. The distribution of differences between the visual and the pupil deficit was compared with a normal distribution using theχ ² goodness-of-fit-test (see Fig. 6 ), but no departure from normality was found (χ²= 0.179 with P > 0.05). 

In general, all patients in this study showed similar results. However, one patient (patient number 15) appeared to be different. The greyscale of his visual field together with the estimates of visual deficit and pupil deficit at each of our five standard locations are shown in Figures 7 A and 7C, respectively. From the greyscale it is apparent that his visual scotoma did not extend beyond the central 5° of the visual field, and indeed his perceptual thresholds were normal at the four standard eccentric locations (Fig. 7C : filled symbols). In contrast, his pupil responses were reduced substantially at these four eccentric locations (Fig. 7C : open symbols), suggesting that in this patient the pupil scotoma extended beyond the visual scotoma. No other patient in this study showed similar results. For comparison, the results from another more typical patient (patient number 14) are shown in Figures 7B and 7D . In this patient, both the visual scotoma and the pupil scotoma extend beyond 17° eccentricity, with the visual deficits exceeding the pupil deficits at all four of the eccentric locations tested. 

By way of summary, the mean estimates (±95% confidence intervals) of visual deficit and pupil deficit in all 19 LHON patients at each of the five tested locations are shown in Figure 8 . The mean visual deficit (filled circles) significantly exceeded the mean pupil deficit (open circles) at all locations (P < 0.001; Wilcoxon signed-ranks test). The average difference between the visual deficit and the pupil deficit was 7.5 dB (95% confidence intervals, = 5.5–9.5 dB) and did not vary significantly between locations (range, 5.2–9.6 dB). 

Discussion

Previous studies ^{2 3 4 5 7} assessing pupil function in LHON have relied on large or full-field light stimuli to elicit the reflex, but the visual deficit in LHON patients is restricted to the central 15° to 30°, with a relatively intact peripheral field. In the present study the pupil light reflex was elicited using stimuli the size of a Goldmann V target presented at five discreet locations within the central scotoma. Light scatter will inevitably have caused some spread of this stimulus, but all the patients in this study were young, with negligible media opacities, and light scatter is likely to have affected patients and control subjects to a similar extent. When tested in this way, the pupil responses were significantly smaller than those found in normal control subjects. Eccentric fixation in some of the LHON patients may explain why stimuli presented in the center produced poor pupil responses. However, eccentric fixation is unlikely to account for a 65% reduction in response size, given that in normal control subjects there is less than 46% difference in the pupil response sizes to standard intensity stimuli presented at fixation and at 17° eccentricity (see Fig. 2 ). Moreover, fixation uncertainty does not explain the poor pupil responses when stimulating peripheral locations, where relatively large changes in stimulus location make little difference to the amplitude of the pupil response. We therefore conclude that there is a significant pupil afferent defect within the visual scotoma in LHON patients. 

A direct comparison of the visual and pupil defects is not possible because different measures of function have been used; visual function has been defined as the light intensity at which the stimulus is perceived 50% of the time, whereas pupil function has been defined as the response amplitude after a standard intensity suprathreshold stimulus. One approach has been to compare the rates of pupil response and perceptual response to perithreshold stimuli in both study groups. This approach is unavoidably subjective. Nevertheless, the results suggest that a stimulus that is below the perceptual threshold is more than three times more likely to be followed by a pupil response in a LHON patient than in a normal control subject, a difference that is unlikely to have arisen by chance (P < 0.001). 

A different approach has been to quantify the pupil defect in terms that are more directly comparable to the measurement of visual defect. Once again, eccentric fixation in some LHON patients will have affected the estimates of both visual and pupil defect. However, we looked carefully at the data but did not find evidence to suggest that patients adopted different fixation strategies in the two tests; any error due to eccentric fixation is therefore expected to have affected both estimates similarly and may not adversely prejudice their comparison. A more systematic difference between the pupil and visual deficit estimates was generated by using different stimulus parameters for the two tests. It is conceivable that “pupil-sparing” was observed, because the larger target used in the pupil test (Goldmann V) extended beyond the limits of the central scotoma, stimulating functioning areas of the visual field not reached by the smaller target used in the visual test (Goldmann III). We feel that this is unlikely to account for our results for two reasons. Firstly, the reverse situation (i.e., the visual test target lying just outside the edge of the scotoma but the larger pupil test target overlapping the edge of the scotoma) is just as likely to have occurred by chance and would have led to just as many instances of the pupil deficit exceeding the visual deficit. Second, the visual scotoma in our LHON patients was uniformly deep and extended well beyond the central 17° tested in this study, making it unlikely that a Goldmann V target would have reached less affected areas of the visual field not stimulated by a Goldmann III target. 

The results of both nonparametric and quantitative analyses in our study suggest that within the central scotoma in LHON the visual deficit exceeded the pupil deficit. This general result was found in all cases except one patient (patient number 15: see Fig. 7 ). His results appear different from the rest of the study group and are contrasted with those of a more typical patient (patient number 14) in Figure 7 . Both patients have the same mutation (11,778) and lost vision about 12 years ago. Of interest, patient number 15 started smoking and drinking heavily 2 years before the onset of his visual loss, whereas MC has never smoked and drinks only in moderation. We cannot say whether excessive alcohol or tobacco consumption influenced the results in patient number 15 or whether these results represent phenotypic variation in LHON. In all the other LHON patients included in this study and at all locations tested, we found relative sparing of pupil function compared with visual function. We did not examine any patients with visual loss of less than 2 years standing and so cannot comment on whether pupillovisual dissociation is present in the acute stages of LHON. 

Jacobson et al. ⁷ were unable to find evidence of pupillovisual dissociation in their recent study of 10 cases with unilateral visual loss. They measured the size of the RAPD using neutral density filters and then compared this with the RAPD expected on the basis of the visual field results using published templates. ^{9 10} However, there is considerable scatter in the data used to establish these published templates for both kinetic ⁹ and automated perimetry ^{10 11} : we would not expect any comparison of visual and pupil function using these templates and the swinging flashlight-test to be sensitive enough to detect the relatively small difference found between visual and pupil deficits in this study (mean, 7.5 dB). 

How are we to interpret “pupil-sparing” in LHON? Anatomic studies in both cats ¹² and primates ¹³ suggest that the pupil light reflex is mediated by W-cells in the retina rather than the X- and Y-cells mediating visual perception. W-cells comprise approximately 10% of all fibers in the optic nerve. They differ from X- and Y-cells in having smaller diameter axons, slower conduction velocities, and lower discharge rates under constant luminance conditions. ¹² The receptive fields of W-cells are similar in size to that of Y-cells, ¹² and so it is unlikely that the centrally placed stimuli used in this study generated pupil responses mediated by peripherally situated ganglion cells. If similar anatomic and neurophysiological differences exist between retinotectal and retinogeniculate fibers in humans, then pupil sparing in LHON may reflect a lower susceptibility of the pupil afferent fibers to this disease than that of the visual afferent fibers. ⁷ Two recent reports lend support to this hypothesis. First, the photic blink reflex, another retinotectal function thought to be mediated by W-cells, also appears to be relatively preserved in LHON. ¹⁴ Second, a histopathologic study of the optic nerve in a patient with LHON showed selective loss of larger diameter fibers with relative preservation of small diameter axons. ¹⁵  

We are not aware of any other examples of pupillovisual dissociation in the literature. However, there are few published studies that have addressed this issue specifically in respect of other optic neuropathies. ¹⁶ It may be that the pupil sparing found in our study is not a distinctive feature of LHON but also can be observed in other diseases of the optic nerve when tested using similar techniques. Of interest, in all three published studies correlating visual field loss and the size of the RAPD, ^{9 10 11} the slopes of the regression lines are less than 1.0, implying that visual loss usually exceeds pupil loss, with some evidence that the slope may vary according to the type of optic nerve disease. ¹⁰ If the relationship between visual deficit and pupil deficit depends on the type of disease, then a study of pupil function in different optic neuropathies may have diagnostic potential. Moreover, the relative susceptibility of pupil afferent fibers compared with visual afferent fibers in different diseases has important implications regarding the underlying mechanisms of damage. 

 Supported by The Joseph Levy Foundation, The T.F.C. Frost Charitable Foundation (FDB), and The Iris Fund (JSH).

 Submitted for publication November 25, 1998; revised March 24, 1999; accepted May 20, 1999.

 Commercial relationships policy: N.

 Corresponding author: F. D. Bremner, Department of Neuro-ophthalmology, National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK.

 

Figure 1.

 

Visual sensitivity in LHON patients relative to normal subjects. Graph shows the mean decibel difference (±95% confidence intervals) between the normal perceptual threshold and that measured in LHON patients at five stimulus locations. Locations tested were fixation (F) and at 17° eccentricity along the 45°/135° meridians in the superotemporal (ST), inferotemporal (IT), superonasal (SN), and inferonasal (IN) quadrants. Normal visual sensitivity appears as 0 dB on this graph. In LHON patients the visual sensitivity was reduced significantly at all five locations, with most deficit at fixation and less at the four eccentric locations.

Figure 2.

 

Pupil responses in LHON patients compared with normal control subjects. Graph shows the mean amplitude (±95% confidence intervals) of the pupil response to a standard intensity (4000 asb) light stimulus presented at each of the five locations. Ordinate values are scaled in percentage of constriction of the pupil area. Results are shown for LHON patients (filled circles) and age-matched normal control subjects (open circles). At all five locations the pupil responses were significantly smaller in the LHON patients, with the greatest difference being at fixation and less difference at the four eccentric locations.

Figure 3.

 

Comparison of pupil and perceptual responses in a LHON patient. Standard intensity (4000 asb) light stimuli were repeatedly presented (n = 14) at 17° eccentricity in the superotemporal quadrant. The chronological order of the stimulus presentations is shown along the abscissa. Perceptual responses to these stimuli are shown as squares (open, perceived; filled, not perceived). Pupil responses to these stimuli are shown as circles (open, response present; filled, no response), and the amplitudes of these pupil responses are shown on the ordinate scale. None of the stimuli were perceived by this patient, but all were followed by a pupil response. The mean amplitude (±95% confidence intervals) of these pupil responses is shown to the right of the axis breaks.

Figure 4.

 

Relationship between stimulus intensity and the size of the pupil response. Graph shows variation in the mean (±95% confidence intervals) pupil response amplitude with stimulus intensity in normal control subjects; all stimuli were presented at fixation.

Figure 5.

 

Quantitative comparison of visual and pupil function in LHON. Graph shows a scatterplot of the estimates of visual deficit (abscissa) and pupil deficit (ordinate) at all five tested locations in all 19 LHON patients (n = 95). The diagonal line represents values of visual deficit equal to pupil deficit; most of the data lies below this line, implying that, in general, the visual deficits exceeded the pupil deficits. The data from patient 15 is shown in filled circles.

Figure 6.

 

Distribution of differences between visual deficit and pupil deficit in LHON. The differences (X) between the visual deficits and the pupil deficits have undergone a simple linear transformation to produce z-scores (z = X − μ/ς, where μ is the mean difference andς is the SD). The distribution of these z-scores appears unimodal and closely matches the standard normal curve shown (dashed line). A χ² goodness-of-fit test confirms that these data do not depart significantly from a normal distribution (χ² = 0.179, P > 0.05).

Figure 7.

 

Visual and pupil function in two patients with LHON. (A, B) Greyscale representations of their visual fields. (C, D) Mean estimates of the visual deficits (filled circles) and the pupil deficits (open circles) at the five stimulus locations tested. (A, C) Results from an exceptional case (patient number 15) in which the central scotoma does not extend beyond the central 5°. Perceptual thresholds were normal at all four eccentric locations tested, and yet the pupil responses were reduced significantly, suggesting that in this case the pupil scotoma extended beyond the visual scotoma. (B, D) Results from a‘ typical’ case (patient number 14), showing a large central scotoma with the visual deficits exceeding the pupil deficits at all four eccentric locations.

Figure 8.

 

Summary of visual and pupil function in LHON. Mean estimates (±95% confidence intervals) of the visual deficit (filled circles) and the pupil deficit (open circles) at each stimulus location. At all five locations, the visual deficit significantly exceeded the pupil deficit (P < 0.05; Wilcoxon signed-ranks test) with an average difference of 7.5 dB.

The pupil perimetry equipment was purchased with a grant from the Joseph Levy Foundation.