Journal of Neuro-Ophthalmology, December 2007, Volume 27, Issue 4

The Prevalence of Relative Afferent Pupillary Defects in Normal Subjects

BACKGROUND: Observational and pupillographic studies of small numbers of normal subjects have shown that a small (<0.3 log units) relative afferent pupil defect (RAPD) is present in a minority. We have extended the investigation of the prevalence of RAPD to a larger number of normal subjects. METHODS: A total of 102 subjects were examined by observation and pupillography. The swinging flashlight test was performed using neutral density filters for quantification. During the pupillographic procedure, light-emitting diodes were placed in front of each eye, alternately flashing for 2.5 seconds with a 0.5 second break. A binocular real-time pupillometer recorded the direct and consensual pupillary responses. After artefact detection and removal, the amplitudes of pupillary response were determined and plotted against stimulus intensity. The means of the direct and the consensual responses were used for automated calculation of RAPD. RESULTS: By observation, there was no RAPD in 87 (85%) subjects; there was an RAPD of 0.15 log units in 13 (13%), and an RAPD of 0.3 log units in 2 (2%). By pupillography, there was an RAPD of 0.07 log units in 53 (52%) subjects, an RAPD between 0.08 and 0.22 log units in 43 (42%) subjects, and an RAPD between 0.23 and 0.39 log units in 6 (6%) subjects. CONCLUSIONS: Observation and pupillographic measurements of the swinging light test in a large normal subject cohort has confirmed that an RAPD is present in a small minority but that it does not exceed 0.39 log units. The RAPD in these subjects may be explained by inaccuracy of measurement or by asymmetries in the connections between visual pathways and pretectal nuclei in the midbrain.

ORIGINAL CONTRIBUTION The Prevalence of Relative Afferent Pupillary Defects in Normal Subjects Helmut Wilhelm, MD, Tobias Peters, Holger Ludtke, PhD, and Barbara Wilhelm, MD Background: Observational and pupillographic studies of small numbers of normal subjects have shown that a small « 0 . 3 log units) relative afferent pupil defect ( RAPD) is present in a minority. We have extended the investigation of the prevalence of RAPD to a larger number of normal subjects. Methods: A total of 102 subjects were examined by observation and pupillography. The swinging flash­light test was performed using neutral density filters for quantification. During the pupillographic pro­cedure, light- emitting diodes were placed in front of each eye, alternately flashing for 2.5 seconds with a 0.5 second break. A binocular real- time pupill-ometer recorded the direct and consensual pupillary responses. After artefact detection and removal, the amplitudes of pupillary response were determined and plotted against stimulus intensity. The means of the direct and the consensual responses were used for automated calculation of RAPD. Results: By observation, there was no RAPD in 87 ( 85%) subjects; there was an RAPD of 0.15 log units in 13 ( 13%), and an RAPD of 0.3 log units in 2 ( 2%). By pupillography, there was an RAPD of 0.07 log units in 53 ( 52%) subjects, an RAPD between 0.08 and 0.22 log units in 43 ( 42%) subjects, and an RAPD between 0.23 and 0.39 log units in 6 ( 6%) subjects. Conclusions: Observation and pupillographic meas­urements of the swinging light test in a large normal subject cohort has confirmed that an RAPD is present in a small minority but that it does not exceed 0.39 log units. The RAPD in these subjects may be explained by inaccuracy of measurement or by asymmetries in the connections between visual pathways and pretectal nuclei in the midbrain. (/ Neuro- Ophthalmol 2007; 27: 263- 267) Department of Pathophysiology of Vision and Neuro- ophthalmology, Universitats- Augenklinik, Tubingen, Germany. Disclosure: Three of the authors ( HW, HL, BW) developed the pupillography system used in this study and would benefit financially if one of the instruments were sold. Address correspondence to Prof. Dr. med. Helmut Wilhelm, MD, Universitats- Augenklinik, Department of Pathophysiology of Vision and Neuro- ophthalmology, D 72076 Tubingen, Germany; E- mail: helmut. wilhelm@ med. uni- tuebingen. de A relative afferent pupillary defect ( RAPD) generally indicates a unilateral or asymmetric disorder of the retina or optic nerve ( 1- 5). The question of whether an RAPD may be present in a completely healthy subject has been addressed in observational and pupillographic studies ( 6- 8). In an observational study ( 6) using the swinging flashlight test in 4,208 subjects, an ophthalmologist found that 49 had an RAPD, and in 2 of these no appro­priate pathologic condition could be identified with further workup. However, the criterion for an RAPD in that study was " pupillary escape," or dilation of the pupil on the involved side when the " swinging light" returns to it from the other side. This criterion precludes sensitivity for small RAPDs. Cox ( 9) found pupillary escape in only 1 of 14 patients who were expected to have an RAPD and con­cluded that the initial constriction is a more sensitive criterion ( 10), one that is less influenced by the presence of physiologic anisocoria ( 11). Most clinicians look for dif­ferences in the initial constriction and for pupillary escape. How many normal subjects would have an RAPD by this method of evaluation is not known. In cases of physiologic anisocoria, it may also be necessary to compare direct and consensual responses of the same eye. Contraction anisocoria ( 12), that is differences between direct and consensual responses, might influence the determination of an RAPD. An optimal observational method of determining an RAPD therefore includes deter­mination of both direct and consensual responses. However, in detection of small RAPDs, pupillo­graphic methods are preferable to observation, as the results are not dependent on the examiner's subjective impression. Moreover, an unlimited number of stimulations can be tested and evaluated in the same patient. In a pupillographic study by Kawasaki et al ( 7), the investigators tested for short- term fluctuation of a pupillographically determined RAPD in 10 healthy subjects and found that the variability depended on the number of stimulus pairs applied. The confidence interval was 0.4 log units if 10 stimulus pairs were tested and only 0.1 log units if 100 stimuli pairs were used. With a stimulus configuration most comparable to ours, an RAPD of 0.3 log units was found in 1 subject. In a second pupillographic study, Kawasaki et al ( 8) examined RAPD fluctuation in 17 healthy subjects over 3 years and found a maximal RAPD of 0.3 log units in 7 of J Neuro- Ophthalmol, Vol. 27, No. 4, 2007 263 J Neuro- Ophthalmol, Vol. 27, No. 4, 2007 Wilhelm et al 68 measurements. In a third pupillographic study, Volpe et al ( 13) compared 13 healthy volunteers with patients who had different amounts of RAPD. They could not dis­tinguish normal subjects from patients with RAPDs of 0.3- 0.6 log units. These studies point out the fact that an RAPD of approximately 0.3 log units may be found in normal subjects, but the number of subjects examined in these studies was small. We report the results of examining 102 normal subjects by observation and by pupillography METHODS Subjects A total of 102 normal subjects ( 41 men and 61 women) aged between 12 and 59 years ( median 29 years) took part in the study. They were recruited from the hospital and among friends of the authors and gave written consent. By history and basic ophthalmologic examination, ocular disorders affecting the pupillary light reflex or causing unilateral afferent defects were excluded. For each subject, visual acuity, slit lamp examination, and direct funduscopy of the optic disc and macular region and cover test were performed. Patients with neurologic disorders, severe head trauma, or diseases that could have led to cerebral ischemia were excluded. None of the subjects were using medication that could affect pupillary function. Classic Swinging Flashlight Test The swinging flashlight test was performed on each patient. The criterion applied was the speed and amplitude of the initial pupillary constriction. Pupillary escape was not observed in any of the subjects. After a swinging flashlight test with at least five cycles using an indirect ophthalmoscope as a light source was performed, a 0.15 log unit neutral density filter was sequentially placed in front of each eye. If there was no RAPD visible or if the RAPD appeared always on the side with the filter, an RAPD of 0 was noted. An RAPD of 0.15 log units was noted if an RAPD was seen without the filter and if the RAPD disappeared when the 0.15 log units filter was held in front of the non- RAPD eye. If the RAPD was present even with a 0.15 log units filter before the non- RAPD eye, the test was repeated with a 0.3 log units filter. An RAPD of 0.3 log units was noted if this filter eliminated the RAPD; an RAPD of 0.15 log units was noted if the RAPD was reversed. Testing with superimposition of 0.3 and 0.15 log units filters always showed an RAPD on the side of the filters. If the 0.15 log units filter could not eliminate an RAPD, perimetry using the Tubingen automated perimeter was performed to exclude subclinical optic nerve disease. The same was done when an RAPD of 0.2 log units or more was measured by pupillography. 264 Pupillographic Swinging Flashlight Test ( SWIFT by Ocuserv Tubingen) A binocular pupillographic system based on an infrared video technique by means of a charge- coupled device ( CCD) camera with a time resolution of 25 Hz and a spatial resolution of 0.24 mm was used. Details of this method have been described extensively previously when the system was used to detect RAPDs in patients with optic neuropathies ( 14). The light conditions were mesopic (- 0.5 cd/ m2 background illumination of the walls of the laboratory). The patient's head was placed on a chin rest 80 cm away from the instrument, which was equipped with a telephoto lens. The pupillary light reflex was elicited by means of two arrays of 12 green light- emitting diodes placed in front of each eye and providing a corneal illu­mination of 60- 80 lux depending of the distance between diodes and the patient's eye. Seven circles of measurements were performed with variable light intensities ( Table 1). Each circle comprised 6 stimuli for each eye. Altogether 42 pairs of stimuli were applied in each subject. Each light stimulus lasted 2.5 s followed by a 0.5 s pause. The whole measurement lasted 252 s. We calculated the means of the direct and consensual constrictions. Measurements with artefacts ( blinks) were discarded ( Fig. 1). Thereafter the mean differences in response amplitudes between stimula­tion of the left and right eyes were plotted against the logarithm of the light stimulus difference and fitted linearly ( Fig. 2). The intersection with the x- axis indicated the RAPD in logarithmic units. The system was tested with artificial RAPDs evoked by neutral density filters and could reliably reproduce RAPDs of 0.15 and 0.3 log units. TABLE 1. Stimulus protocol used with the automated swinging flashlight test* Right eye relative intensity 100 100 50 25 100 100 10 Left eye relative intensity 100 50 100 100 25 10 100 Filter density ( log unit) No filter 0.3 left 0.3 right 0.6 right 0.6 left 1.0 left 1.0 right From Wilhelm et al ( 14). * To calibrate the instrument, the brightness of the light- emitting diode arrays was measured exactly. These values usually differed in small amounts from the ideal values shown in this table ( for example, 96 instead of 100). For calculation of the RAPD, those exact values were used rather than the " ideal" values listed in this table. The total corneal illumination at 100% brightness was between 60 and 80 lux depending on the individual orbital anatomy. © 2007 Lippincott Williams & Wilkins Afferent Pupillary Defect J Neuro- Ophthalmol, Vol. 27, No. 4, 2007 2 right £ £ off . g* left rx7 = P^° 1 FIG. 1. Recording of an artificial relative afferent pupillary defect ( RAPD) produced by a 1 log unit gray filter placed before the left eye. This noisy recording shows how artefact management worked. The arrows indicate when a pupillary constriction is to be expected. Some responses remain subthreshold when the left eye is stimulated. No. 1 indicates an artefact, usually a blink, which is recognized because - « - j- jme , c it appears outside the time frame for a pupillary constriction. No. 2 indicates a blink artefact that appears during the normal light response and is superimposed. It would lead to an overestimation of the RAPD and is recognized because the already dilating pupil starts to constrict again ( 17). VZ3 J RESULTS The clinical swinging flashlight revealed no RAPD in 87 ( 85%) of 102 subjects, an RAPD of 0.15 log units in 13 ( 13%) subjects, and an RAPD of 0.3 log units in 2 ( 2%) subjects. Higher RAPDs did not occur. The overall mean of the observationally measured RAPD of all subjects ( independent of the side of the RAPD) was 0.02 log units, 0.07 log units below the pupillographic mean. mm 0.4 I . g) 0.2 ] E 0.0 4 - 0.2 - 0.4 \ • T T •• *?/\ • a / { I i ) L- : x i - 1.0 - 0.5 0.0 0.5 1.0 FIG Illumination difference ( right - left, in log units) 2 Calculation of the relative afferent pupillary defect ( RAPD). Mean response amplitude differences ( right eye stimulation minus left eye stimulation) were calculated and plotted against illumination differences between the two eyes in logarithmic units. Negative means a brighter light before the left eye; positive means a brighter light before the right eye. A linear curve is fitted and the intersection with the abscissa gives the amount of the RAPD. In this example, the right RAPD is + 0.12 log units. Pupillographic results were similar to the clinical findings ( Fig. 3). An RAPD of 0.07 log units or less was present in 53 ( 52%) subjects, between 0.08 and 0.22 log units in 43 ( 42%) subjects, and between 0.23 and 0.39 log units in 6 ( 6%) subjects. The maximum RAPD was 0.39 log units, found in one subject. The mean RAPD was 0.09 log units if the amount of RAPD was taken independently from its laterality. If a right RAPD was calculated as pos­itive and a left RAPD as negative ( Fig. 3), the RAPD was + 0.02 with a SD of ± 0.12. Among our subjects, 95% had results in the range between + 0.25 and 0.25 log units ( pos­itive value means RAPD on right; negative value means RAPD on left). No visual field defect was found pupillo-graphically in the subjects with RAPD > 0.2 log units. There was a weak correlation between the RAPD measured by observation and by pupillography ( R2 = 0.056, P < 0.03) ( Fig. 3). DISCUSSION This study has confirmed the results of former studies on the prevalence of an RAPD in normal subjects. An RAPD of 0.3 log units, the threshold of detectability by the swinging light test performed with good technique, was present in fewer than 2% of 102 subjects by careful observation and pupillography. An RAPD of > 0.39 log units was not found. Whether the prevalence of RAPDs within a normal population older than 60 years would be higher cannot be answered by our study. Short- term fluctuation and inaccuracy of measure­ment might explain a certain variability of the results at RAPDs of 0.15 log units, but it is unlikely that an RAPD of 0.3 log units measured pupillographically would be explained by inaccuracy of measurement alone. Comparing the present study to the first study of Kawasaki et al ( 8), the application of 42 pairs of stimuli in each subject should have produced short- term variability slightly below 0.2 log units. Kawasaki et al ( 8) observed this variability with 40 pairs of stimuli. 265 J Neuro- Ophthalmol, Vol. 27, No. 4, 2007 Wilhelm et al 0.45 0.30 0.15 in o 0.00 a a. - 0.15 - 0.30 - 0.45 • Observation with gray filters D Pupillography D 1Q" mfnmFri11 20 i 40 i 60 i 80 i 100 FIG. 3. The relative afferent pupil­lary defect ( RAPD) for all subjects measured by observation with gray filters and pupillographically ( posi­tive values indicate mean RAPD in the right eye, negative values indicate mean RAPD in the left eye). This figure also illustrates the poor correlation between the results from observation of the swinging flashlight test and its pupillographic measurement. Subjects An RAPD in normal subjects that exceeds 0.15- 0.2 log units is probably an expression of natural fluctuations in the pupillary system. Perhaps it is the afferent counterpart to physiologic anisocoria. In the study of Kawasaki et al ( 8) on long- term RAPD variability, there were subjects who had randomly reversing RAPD between the two eyes, but in 8 of these 17 subjects, the RAPD was always on the same side ( 8). This phenomenon might be due to diffuse asym­metry of retinal light sensitivity. Congenital or acquired subclinical lesions are a second possible explanation. A third explanation may be asymmetry within the connect­ing pathways between the afferent visual system and the pretectal area. Some patients with lesions in the pulvinar-pretectal area show an unequivocal RAPD without any measurable visual deficit ( 15- 18). On the basis of our study and previous reports, an RAPD > 0.3 log units ( 11) is likely to be an expression of a substantial lesion of the afferent pathways and requires further evaluation. An RAPD of < 0.3 log units in an otherwise healthy subject with normal visual fields may be accepted as a normal finding. Pupillographic measurement of the RAPD may be helpful in unclear cases and clinical trials ( 19). It has the advantage of avoiding mistakes that occur in observational methods, especially different light exposure times for the two eyes. Automated evaluation removes examiner bias and the problems related to comparison of examinations of different clinicians. Binocular measurement of the pupils reduces the influence of physiologic and contraction anisocoria on the results. Awareness of these problems motivated the development of an automated procedure and device ( SWIFT by Ocuserv Tubingen) ( 14). The aim was to come close to the clinically applied swinging flashlight test by choosing an illumination time of 2.5 s and an alternating time of 0.5 s. Kawasaki et al ( 7) tested different com­binations of illumination and alternating times and could not demonstrate a clear preference. However, they showed that steps smaller than 0.3 log units did not increase the accuracy of the measurement. Because of its stepwise noncontinuous characteristics, the gray filter method may produce a slight underestimation of the clinical RAPD because results between 0.15 and 0.3 log units are put on 0.15 log units and results between 0 and 0.15 are put on 0 log units. This result is expressed by a lower mean of the observational RAPD than of the pupillographic RAPD. The poor correlation between observationally and pupillographically assessed RAPDs in this study is puz­zling. However, the range of RAPDs found in our subjects was very narrow, and the vast majority of the data points clustered near 0. Therefore, one could hardly expect a good correlation. The two subjects with pupillographically mea­sured RAPDs > 0.3 log units had no RAPD when examined by observation, which may indicate that natural variability plays a major role. 266 © 2007 Lippincott Williams & Wilkins Afferent Pupillary Defect J Neuro- Ophthalmol, Vol. 27, No. 4, 2007 REFERENCES 1. Thompson HS. Afferent pupillary defects: pupillary findings associated with defects of the afferent arm of the pupillary light reflex arc. Am J Ophthalmol 1966; 62: 860- 73. 2. Thompson HS, Watzke RC, Weinstein JM. Pupillary dysfunction in macular disease. Trans Am Ophthalmol Soc 1980; 78: 311- 7. 3. Thompson HS. Pupillary signs in the diagnosis of optic nerve disease. Trans Ophthalmol Soc UK 1976; 96: 377- 81. 4. Folk JC, Thompson HS, Farmer SG, et al. Relative afferent pupillary defect in eyes with retinal detachment. Ophthalmic Surg 1987; 18: 757- 9. 5. Lam BL, Thompson HS. A unilateral cataract produces a relative afferent pupillary defect in the contralateral eye. Ophthalmology 1990; 97: 334- 8. 6. Levatin P, Prasloski PF, Collen MF The swinging flashlight test in multiphasic screening for eye disease. Can J Ophthalmol 1973; 8: 356- 60. 7. Kawasaki A, Moore P, Kardon RH. Variability of the relative afferent pupillary defect. Am J Ophthalmol 1995; 120: 622- 33. 8. Kawasaki A, Moore P, Kardon RH. Long- term fluctuation of relative afferent pupillary defect in subjects with normal visual function. Am J Ophthalmol 1996; 122: 875- 82. 9. Cox TA. Pupillary escape. Neurology 1992; 42: 1271- 3. 10. Cox TA. Initial pupillary constriction in the alternating light test. Am J Ophthalmol 1986; 101: 120- 1. 11. Cox TA. PupiUographic characteristics of simulated relative afferent pupillary defects. Invest Ophthalmol Vis Sci 1989; 30: 1127- 31. 12. Lowenstein O. Alternating contraction anisocoria: a papillary syndrome of the anterior midbrain. Arch. Neurol. Psychiatry 1954; 72: 742- 57. 13. Volpe NJ, Plotkin ES, Maguire MG, et al. Portable pupillography of the swinging flashlight test to detect afferent pupillary defects. Ophthalmology 2000; 107: 1913- 21. 14. Wilhelm B, Liidtke H, Peters T, et al. Automatisierter Swinging-flashlight- Test bei Patienten mit Sehnervenerkrankungen. Klin Monatsbl Augenheilk 2001; 218: 21- 5. 15. Eliott D, Cunningham ET Jr, Miller NR. Fourth nerve paresis and ipsilateral relative afferent pupillary defect without visual sensory disturbance: a sign of contralateral dorsal midbrain disease. J Clin Neuroophthalmol 1991; 11: 169- 72. 16. Ellis CJ. Afferent pupillary defect in pineal region tumour. J Neurol Neurosurg Psychiatry 1984; 47: 739^ 1. 17. Johnson RE, Bell RA. Relative afferent pupillary defect in a lesion of the pretectal afferent pupillary pathway. Can J Ophthalmol 1987; 22: 282^. 18. Staubach F, Pieh C, Maier P, et al. Relative afferent pupillary defect with normal vision and vertical strabismus- implications for pupillary pathway anatomy. Graefes Arch Clin Exp Ophthalmol 2007; 245: 321- 3. 19. Wilhelm B, Liidtke H, Wilhelm H. Efficacy and tolerability of 0.2% brimonidine tartrate for the treatment of acute non- arteritic anterior ischemic optic neuropathy ( NAION): a 3- month, double- masked, randomised, placebo- controlled trial. Graefes Arch Clin Exp Ophthalmol 2006; 244: 551- 8. 267