Entrance Pupil Size Predicts Retinal Illumination in Darkly Pigmented Eyes, But Not Lightly Pigmented Eyes

Abstract

Purpose.: We determined the effect of entrance pupil size on retinal illumination. The influence of unilateral miosis on the magnitude of the pupil light reflex was studied to ascertain how a clinically significant anisocoria influences the relative afferent pupil defect (RAPD).

Methods.: Miosis was induced by topical 1% pilocarpine in the right eye of 14 healthy subjects with normal eyes. The interocular difference in retinal illumination was assessed by computerized pupillometry from the stimulus response curve of the right and left eyes. The main outcome measure was the RAPD, determined by computerized pupillography, at baseline and after pilocarpine-induced anisocoria.

Results.: Induced anisocoria produced a significant change in RAPD from baseline (mean = 1.60 dB in the miotic eye, P = 0.007). However, anisocoria correlated with RAPD only in subjects with darkly pigmented irides (Pearson correlation coefficient 0.793, P = 0.05).

Conclusions.: In darkly pigmented eyes, entrance pupil size significantly influenced the retinal illumination. However, retinal illumination of lightly pigmented eyes is relatively independent of entrance pupil size, presumably due to extrapupillary transmission of light through the iris and sclera. This has important implications in understanding the potential influence of anisocoria on the RAPD and also greater susceptibility of lightly pigmented eyes to light toxicity.

Introduction

The primary function of the pupil is to modulate retinal illumination. ¹ Based on this concept and assuming equal retinal adaptation, unequal pupil size (anisocoria) would be expected to create unequal retinal illumination between the two eyes. In addition to the effect of pupil size on the amount of light entering the eye, pupil size also may influence the effectiveness of photoreceptor function based on the angle of light rays with respect to the photoreceptors, also known as the Stiles-Crawford effect. ²  

In the clinical setting, unequal retinal illumination between the two eyes can be observed as a relative afferent pupillary defect (RAPD) and typically the cause is a pathologic lesion of the optic nerve or retina. Because asymmetric entrance pupil size may create unequal retinal illumination as well, this may increase or decrease artificially the estimate on a pathology-related RAPD. ³ Pupil size also has been considered an important factor in other clinical tests, such as the scotopic threshold test for glaucoma ^4,5 and the Ganzfeld ERG test. ^6,7 In addition, the role of pupil size is also important in understanding the cumulative pathologic effects of light on the internal structures of the eye in terms of cataract formation ^8,9 and retinal degeneration. ^10–20 Therefore, in any study of light on the eye, including its role in disease, understanding the role of pupil size on effective retinal illumination has great importance. ^21–23  

The purpose of our study was to understand better the effect of asymmetric entrance pupil size (anisocoria) on retinal illumination and function using computerized pupillometry. With our instrument, both pupils could be recorded simultaneously while stimulating either the right or left eye to estimate the response function over a large range of stimulus light intensities. The pupil light reflex was used in this study because it is an objective, biological light meter whose magnitude reflects effective retinal illumination. ^24–26  

Methods

We tested by computerized pupillometry 14 healthy subjects with no history of ocular disease or trauma, after informed consent was obtained according to the tenets of the Declaration of Helsinki. Approval for this study was obtained through the Institutional Review Board of the University of Iowa. Each subject was tested at baseline and at 90 minutes after the instillation of 1% pilocarpine into the right eye. Between testings, the subjects were instructed to maintain the same level of adaption to the indoor lighting level of the ophthalmology clinic. 

Pupil responses to light stimuli were recorded using a computerized infrared pupillometer (Visual Pathways, Inc., Prescott, AZ) which presented a 30° radius light stimulus to each eye in non-Maxwellian view. Each subject was adapted to a 3.1 apostilb background light for 30 seconds, then a sequence of light stimuli was presented alternately to each eye at varying intensities above the background level. During the test, the stimulated eye was allowed to foveate on a small central cross before the ensuing light stimulus to control fixation. The refractive power of the instrument was adjusted by entering the refraction of the subject into the instrument's software so that the focal point was set at infinity to control accommodation. A flat black metal septum separated the right and left eye optical pathways to minimize stray light scatter. The stimulus duration was 0.2 seconds and the time between light stimuli was 3.3 seconds. The stimulus duration was well within the latency time of the pupil light reflex so that the entrance pupil size was not affected during the time that the stimulus was on. Eight stimulus intensities were presented over a 3.5 log unit range (35 dB range; −44 to −9 dB of attenuation above a background of 3.1 apostilbs; −44 dB attenuation = 0.13 cd/m², −0.9 dB = 400.7 cd/m²). Each stimulus was repeated six times during the test in staggered fashion. The order of the stimuli was from dim to brightest stimuli for the stimulus protocol, the right eye received the same intensity as the left eye for each intensity step. Each subject was tested with the same protocol. The order of the stimuli was the same for each subject. Previous unpublished studies demonstrated that reversal of the stimulus order has no significant effect on the fitted response function due to retinal adaptation. Because of the induced miosis after instillation of pilocarpine, only the movement of the untreated left pupil was recorded and analyzed (sampling rate = 60 Hz) at the 90-minute posttreatment test time. The pupil diameter of the treated right eye was recorded before and after pilocarpine using a magnified infrared video camera with millimeter rule in the picture so that the entrance pupil size could be measured in all cases, even after profound miosis. This was necessary because at very small pupil sizes (<2.0 mm diameter), the computerized infrared pupillometer was not always able to track and measure the bright (infrared retro-illuminated) image of the pupil. 

Pupillographic tracings were stored and analyzed (Fig. 1) using a software program described in a previous report. ²⁷ The effective retinal illumination was estimated from the stimulus response curve (fit by the Naka Rushton equation) ^28–30 based on the pupil light reflexes recorded at different intensities, at baseline, and after induction of unilateral miosis. The RAPD was determined at an intensity of 20 dB of attenuation (100 apostilbs or 31.8 cd/m² above a background level of 3.1 apostilbs). The RAPD was calculated as the interocular difference of the stimulus response curves at this intensity (Fig. 2). The effective retinal illumination over the entire range of stimulus intensities was assessed as the area under the stimulus response curve and the afferent asymmetry between the two eyes was estimated from the area difference between the two eyes (Fig. 2). 

Figure 1. 

 

Example of how a miotic pupil may decrease the amplitude of the pupil light reflex by reducing entrance of light through the smaller pupil opening. The stimulus light is alternated between the right and left eyes, and the movement of the left pupil is recorded as shown by the tracings. The beginning and peak of the pupil tracing are marked on the tracings with a thin vertical line by the software analysis program. Time (in seconds) is shown along with the square wave stimulus. In the baseline state before pilocarpine (top), the pupils are equal and the pupil light reflexes are of the same amplitude in response to right or left eye stimulation. At 90 minutes after 1% pilocarpine to the right eye (bottom), the pupil is miotic and immobile, and the left, untreated pupil responses are smaller with right eye stimulation compared to left eye stimulation, causing an RAPD.

Figure 2

 

Example of stimulus response curves plotted for baseline state before pilocaripin (top) and after miosis (bottom) was induced in the right eye by pilocarpine (same patient as example in Fig. 1). The stimulus response curve is shown for the left, untreated eye's pupil contraction amplitude when the right eye was stimulated (as a function of light intensity) and also for the left eye's pupil contraction for left eye stimulation. As the stimulus light was made brighter, the pupil contracted more, producing a sigmoid-shaped response curve (see Methods). The area under this curve (shaded) essentially was the same no matter which eye was stimulated in the baseline state, as shown in the top graph. When the right eye was made miotic, the area under the stimulus-response curve decreased compared to pretreatment and compared to the fellow eye stimulation. The lightly shaded area in the bottom graph reflects the decreased response of the stimulated right eye due to the reduced entrance size of the pupil. The RAPD was estimated by quantifying the intensity difference between the two curves depicted by the distance between the vertical dashed lines (in this case at the stimulus intensity level of 20 dB on the stimulus-response curve of the untreated eye). For example, in the bottom graph, a decrease of intensity of 2 dB (0.2 log units) was needed to diminish the pupil reaction from the normal eye to produce a contraction equal to that produced by stimulation of the treated eye with the miotic pupil.

Because the effect of anisocoria on the RAPD seemed to vary widely among the subjects during the initial analysis, one of the investigators (AK) was asked to rank each subject by the degree of ocular pigmentation based on clinical observation of the iris color of each subject. This investigator was not told the reason for this at the time, so as to maintain an unbiased estimate of iris color. The subjects were ranked in order of increasing iris pigmentation from 1 to 14. Since all of the subjects were available in the immediate vicinity of the study location, their ranking by iris pigmentation was done after the testing of all subjects was completed by visual inspection of their iris color during the same day by memory. Iris pigmentation was correlated with the change in RAPD, change in the area difference under the stimulus-response curves of the two eyes, and the degree of induced anisocoria using a Spearman rank correlation test. 

For each patient, the entrance pupil size and degree of anisocoria were tabulated at the light intensity of 20 dB attenuation, the intensity at which the RAPD was measured (see above). The expected retinal illumination (in dB) of each eye was derived mathematically based on the entrance pupil size measured at each intensity of stimulus from the following equation: Retinal illumination (in log units) = log (area of pupil) + log (stimulus intensity) under conditions of clear ocular media and a standardized axial eye length. ^31–33 Since retinal illumination is being calculated in log units, log retinal illumination = log (pupil area times stimulus intensity). The expected retinal illumination was calculated with and without correction for the Stiles-Crawford effect, ^2,34 which takes into account the effect of pupil size on the response of the photoreceptors, as influenced by the angle of light rays entering the pupil with respect to the orientation of the photoreceptors. This effect is not apparent when pupil diameters are 6 mm or less and did not change the calculations significantly for the “expected RAPD” over the range of pupil sizes studied in our experiments. The expected RAPD (in dB units) was the difference of the calculated retinal illumination between the two eyes for a measured amount of anisocoria. 

Two of the subjects (one darkly pigmented, subject 14, and one lightly pigmented, subject 1) also were tested at various degrees of miosis by using increasing concentrations of pilocarpine in the right eye on the same day (1/8%, 1/4%, 1/2%, 1%). This was done to establish the relationship between pupil size and effective retinal illumination within a given subject (as measured by the RAPD) to compare two subjects with widely different degrees of iris pigmentation. 

Results

An example of pupil light reflexes recorded from the mobile, untreated pupil of one subject with darkly pigmented irides is shown in Figure 1. In the figure, the subject's pupil light reflexes to an alternating light stimulus are shown before (top) and after (bottom) the production of miosis. The pharmacologically constricted pupil caused a smaller amplitude pupil light reflex when the miotic right eye was stimulated compared to the untreated left eye. The results from the same subject are shown in Figure 2, in which the amplitude of the left (untreated) pupil light reflex was plotted as a function of the stimulus light intensity for the right and left eye stimulation. Before the right eye was treated with pilocarpine, there was no difference in the area under the stimulus response curves for right and left eye stimulation. After miosis was induced in the right eye, the stimulus response curve from right eye stimulation fell short of the left eye's stimulus response curve at all intensities and, hence, the area under the curve was less. This result was anticipated because of the difference in entrance pupil size that was induced between the two eyes; however, as will be reported, this was not found in all of the subjects. 

The predicted retinal luminance for each subject was calculated based on the stimulus light intensity and pupil area at the time of the stimulus onset (see Methods). The calculated difference in retinal luminance (expected RAPD) was plotted as a function of the degree of anisocoria produced by pilocarpine in each of the 14 subjects (Fig. 3). There was a significant linear correlation between the anisocoria produced and the expected RAPD (correlation coefficient r ² = 0.92). The correlation was not a perfect line for the expected RAPD. This was because the effect of anisocoria on retinal illumination also is dependent upon the actual pupil size at a given level of anisocoria. That is, 1 mm anisocoria for a 5 and 6 mm pupil of a subject would not have the same effect as a 3 and 4 mm pupil in the same subject, because a difference in pupil size based on diameter measurements does not reflect the difference in area, which increases as pupillary diameter increases for any given level of anisocoria. In this case, the area difference was greater for a subject with the 5 and 6 mm pupils compared to a subject with the 3 and 4 mm pupils. 

Figure 3. 

 

The relationship between the anisocoria produced by pilocarpine in the 14 subjects and the expected RAPD, predicted by the difference in retinal luminance calculated from the difference in pupil entrance size. The high linear correlation is anticipated, because the expected RAPD values are based on mathematical calculations. The correlation was not a perfect line for the expected RAPD because the effect of anisocoria on retinal illumination also is dependent upon the pupil size (area) at a given level of anisocoria (see Results).

The Table summarizes the effects of unilateral miosis on the difference in RAPD between the baseline condition and after pharmacologic miosis. As seen from the mean values at the bottom of the Table, there was an overall increase in the RAPD of 1.60 dB (0.160 log units) following induced miosis (P = 0.007, mean anisocoria was 2.2 mm). However, there did not seem to be a consistent effect of anisocoria on the RAPD among subjects. In some subjects, there was a definite decrease in the pupil light reflex in the eye with miotic pupil, thus producing an RAPD. However, in other subjects, very little or no RAPD was produced, even though a significant miosis (anisocoria) was present. 

Table.

 

Summary of Characteristics and Pupil Data From Normal Subjects Before and After Receiving 1% Pilocarpine in the Right Eye

Subject No. Ranked by Iris Pigment From Least Pigmented to Most Pigmented	Baseline Pupil Size, mm		Anisocoria, mm Baseline OS–OD	Baseline RAPD,* dB	Post-Pilocarpine Pupil Size, mm		Anisocoria, mm, Post- Pilocarpine OS–OD	Post- Pilocarpine RAPD,* dB	Change in RAPD Post Pilocarpine- Baseline, dB
	OD	OS			OD	OS
1	4.76	4.62	−0.14	1.44	2.21	5.26	3.05	3.3	1.86
2	5.33	4.68	−0.65	3.13	4.00	4.96	0.96	3.03	−0.1
3	5.45	5.36	−0.09	4.76	4.00	5.37	1.37	4.9	0.14
4	4.47	4.44	−0.03	−0.95	2.00	4.71	2.71	−0.71	0.24
5	3.68	3.60	−0.08	3.31	3.00	4.47	1.47	5.1	1.79
6	5.55	4.56	−0.99	1.68	5.00	5.32	0.32	3.28	1.6
7	5.42	5.54	0.12	−0.42	2.00	5.55	3.55	−2.28	−1.86
8	4.85	4.51	−0.34	3.22	2.50	4.51	2.01	4.15	0.93
9	6.46	6.12	−0.34	−4.32	4.50	6.15	1.65	−0.38	3.94
1	4.91	4.69	−0.22	−1.06	3.00	4.85	1.85	3.74	4.8
11	5.31	5.73	0.42	0.43	2.50	5.73	3.23	1.08	0.65
12	6.06	5.71	−0.35	−0.40	2.00	6.06	4.06	2.17	2.57
13	4.95	4.65	−0.30	−4.10	2.50	4.78	2.28	−2.66	1.44
14	5.99	6.00	0.01	−0.12	3.74	5.99	2.25	4.22	4.34
Mean	5.23	5.01	−0.21	0.47	3.07	5.27	2.20	2.07	1.60
SD	0.71	0.73	0.34	2.68	1.00	0.57	1.04	2.61	1.85

*  RAPD in dB (positive value is a right RAPD, negative value is a left RAPD); log units = dB/10. RAPD was determined by two different methods using computerized pupillography. The RAPD values in this Table were determined by determining the dB attenuation needed in the more responsive eye to reduce its pupil response to that of the opposite eye (see Methods and Fig. 2).

To try to understand this discrepancy, we plotted a scattergram of the relationship between the pharmacologically-induced anisocoria and the change of measured RAPD from baseline, as shown in Figure 4. Unexpectedly, there was no significant linear correlation (P = 0.84). The lack of correlation between anisocoria and change of RAPD was not anticipated and, accordingly, a reason was sought for this result. 

Figure 4. 

 

This scattergram shows the lack of a relationship between the anisocoria produced by pilocarpine and the change of measured RAPD derived from computerized pupillography in the same 14 subjects shown in Figure 3. The method of RAPD determination for the values used in this graph consisted of determining the dB attenuation needed in the more responsive eye to reduce its pupil response to that of the opposite eye (see Methods and Fig. 2).

One possibility that was considered was that retinal illumination during the light stimulation might not have been solely determined by pupil size, but also by light penetration through the wall of the eye. It was hypothesized that the extrapupillary pathway for light entry into the eye would be more significant in subjects with less ocular pigmentation, regardless of pupil entry size. This hypothesis was explored by first ranking the subjects according to iris pigmentation, from least to greatest by one of the investigators (AK), who was masked to the hypothesis when asked to rank the subjects. Rank iris pigmentation then was correlated with the degree of effective retinal illumination as determined by pupil responses between the treated and untreated eye that was induced by miosis. 

Iris pigmentation rank did not correlate significantly with the change in RAPD after induced miosis in the 14 subjects (correlation coefficient r = 0.169, P = 0.552), nor was there a significant correlation between pigment rank and degree of miosis produced by pilocarpine (r = 0.359, P = 0.201). 

A correlation between anisocoria and the change of the interocular difference of the area under the curve from baseline was replotted; the most darkly pigmented eyes (rank 9–14) were plotted separately from the lightly pigmented eyes (rank 1–6), as shown in Figure 5. A significant correlation was found for the more darkly pigmented eyes (Pearson correlation coefficient 0.793, P = 0.05), but not for the lightly pigmented eyes (Pearson correlation coefficient 0.112, P = 0.8). 

Figure 5

 

Anisocoria versus RAPD is replotted using change from baseline of the interocular area under the stimulus–response curve after pilocarpine as a measure of RAPD. In addition, the subjects were divided equally into two groups based on their rank of pigmentation; those with lightly pigmented eyes (rank 1–6) and those with darkly pigmented eyes (rank 9–14). With this subdivision, there was a significant linear correlation for the subjects with darkly pigmented eyes (closed circles), but no correlation for the subjects with lightly pigmented eyes (open circles).

Because of the apparent correlation between ocular pigmentation and effective retinal illumination, we explored this concept even further in two subjects, one of whom was darkly pigmented (subject 14) and the other who was lightly pigmented (subject 1). We hypothesized that in the darkly pigmented subject, pupil entrance size would have a greater effect on retinal illumination and in the lightly pigmented subject it would have a much lesser effect. To confirm this, we repeated the initial experiment using increasing concentrations of pilocarpine (1/8%, 1/4%, 1/2%, and 1%) to the right eye on the same day, starting with the most dilute solution. This enabled us to ascertain the effect of varying degrees of miosis (and, hence, anisocoria) on retinal illumination. The results of this experiment are shown in Figure 6. Two linear correlations are shown for each subject. In one correlation, the measured change in RAPD is correlated with anisocoria. In the second correlation, the expected change in RAPD that would be predicted by the entrance pupil size (see Methods) was calculated and graphed. The measured change in RAPD in the darkly pigmented subject was almost the same as that predicted from the calculations according to entrance pupil size that was recorded, and the linear correlations were nearly identical. In contrast, for the lightly pigmented subject there was almost no effect of anisocoria on the measured change in RAPD. 

Figure 6

 

(A) Linear correlation between anisocoria and RAPD change in the most darkly pigmented subject (subject 14) who had different degrees of anisocoria induced by increasing concentrations of pilocarpine on the same day. The RAPD that actually was measured showed a very high correlation with the amount of anisocoria and this line (broken line, R ² = 0.7206) was almost exactly the same as the RAPD that would have been expected based on mathematical calculations of pupil entrance area (solid line, R ² = 0.9828). (B) Similar graph as (A), but now for the most lightly pigmented subject (subject 1). In contrast to (A), this subject's change in RAPD over different degrees of anisocoria induced by increasing concentrations of pilocarpine showed no relationship to the degree of anisocoria (broken line, R ² = 0.0959). The line was almost flat compared to the line of expected RAPD change based on calculations (solid line, R ² = 0.9906).

Discussion

Our study originally was conceived to understand better the relationship between asymmetric pupil size and the relative afferent pupillary defect. The majority of previous studies have used psychophysical means of assessing retinal illumination and light scatter based on how entrance pupil size and ocular pigmentation affect the perception of light in terms of brightness or light flicker. ^35,36 The effect of entrance pupil size on the pupillary light reflex provides an alternative means of measuring effective retinal illumination in an objective manner. In some clinical situations, a unilateral mydriasis or miosis of the pupil may influence the clinical estimate of the RAPD. Depending on which pupil is larger or smaller, the RAPD can be underestimated or overestimated. One report studied the effect of unilateral pharmacologic mydriasis on the log unit RAPD determined clinically using neutral density filters. ³ The results showed that unilateral dilation of the pupil in normal subjects may induce an RAPD in the opposite eye, but the relationship between amount of induced anisocoria and the magnitude of the RAPD was not as highly correlated as was to be expected. In a subset of our study, patients with an existing RAPD had either the affected or unaffected eye dilated pharmacologically. The induced anisocoria did not produce a predictable effect on the RAPD in these patients, which was difficult to explain. 

Because one pupil was fixed by a miotic agent, the untreated pupil was recorded during the alternating light test to determine the RAPD. It is known that some normal subjects have a greater direct than consensual pupil contraction, depending on which eye is stimulated, termed contraction anisocoria. Although contraction anisocoria theoretically may add or subtract from the calculated RAPD if only one pupil is recorded, the effect does not change in a given person over time. In the context of our study, a contraction anisocoria, if present, would not affect the net change in RAPD induced by changing the entry pupil size in one eye. During baseline recording of both pupils before administration of the topical miotic agent, we did not find, in fact, any significant contraction anisocoria in the normal subjects tested in this study. Since the pupil light reflex is a built-in objective light meter of the eye, we chose to reexamine the relationship in a very controlled way by recording entrance pupil size carefully and measuring the resulting effective retinal illumination using the amplitude of the pupil light reflex. By using a computerized pupillometer to provide a controlled stimulus and a precise recording of pupil dynamic behavior, we felt that it would be possible to confirm a relationship between anisocoria and effective retinal illumination, if one existed. Care was taken in the experimental design to control for the state of adaptation, and delivery of a repeatable and accurate light stimulus over a range of intensities that could be given in an alternating fashion to the right and left eye. A short duration light stimulus was given (within the latency time of the pupil light reflex) so that entrance pupil area would not be affected by the light stimulus. 

The Stiles-Crawford effect was found to be minor in the context of our study. This is due to a number of reasons. First, the Stiles-Crawford effect is greatest with large sized pupils and reducing pupil size by a miotic tends to minimize this effect. Second, the stimulus area of the retina is quite large and is even greater with brighter light stimuli due to light scatter, and not only would recruit rods (although rods do not contribute greatly to the Stiles-Crawford effect, since it occurs mainly under photopic conditions), but also would recruit contributions to the pupil light reflex from intrinsic activation of melanopsin containing retinal ganglion cells, which would not be affected by the Stiles-Crawford effect, since these neurons are not known to have directional sensitivity. 

Among the 14 normal subjects tested, it was easy to demonstrate that on the average, an induced pharmacologic miosis in one eye produced a relative decrease in retinal illumination in the treated eye compared to the fellow, untreated eye. However, we were surprised to find that, overall, there was no significant relationship between anisocoria and retinal illumination. A recent study has shown that with bright blue light (but not with red light), the postillumination sustained pupil contraction was reduced by a smaller pupil size, but iris color was not specified. ³⁷ The initial contraction amplitude was unaffected by pupil entrance size, which may have resulted from the nonlinearity in pupil response at high intensity. Why were the RAPD results in this study so variable among the normal subjects for the same degree of anisocoria under the same carefully controlled conditions? 

We postulated that in lightly pigmented subjects an additional pathway exists for retinal illumination besides light entry through the pupil. Such an “extrapupillary” light pathway would allow light to penetrate the sclera, uvea, and/or iris. We surmised that penetration through the eye to the retina would be limited partly by the degree of melanin pigment in the eye. It would follow that large amounts of melanin pigmentation would constrain most of the light to enter through the pupil and, therefore, darkly pigmented subjects would show the most effect of pupil size on retinal illumination. Conversely, lightly pigmented subjects would allow light to pass through the sclera and uvea, and hence, would be expected to show much less dependence of retinal illumination on entrance pupil size. This was confirmed by pharmacologically varying the entrance pupil size over a wide range in two of the subjects, one darkly pigmented and the other lightly pigmented. We found that the measured change in RAPD due to anisocoria was nearly exactly that which was predicted in the darkly pigmented subject, indicating that entrance pupil size was the main, if not sole, determinant of the amount of effective retinal illumination. In contrast, the lightly pigmented subject showed no effect of anisocoria on the RAPD, indicating that light can penetrate the eye through extrapupillary pathways when there is not sufficient ocular pigmentation to absorb it. 

Although this concept seems obvious, we believe that it may be more difficult to demonstrate using standard psychophysical testing, since subjective light perception may not be as accurate as using the pupil light reflex. In one study of ocular scatter of light, it also was found that a significant degree of light presented as an annulus outside the confines of the pupil could reach the retina and produce scatter in lightly pigmented subjects. ³⁶ The pupil light reflex is capable of summating the area of retina stimulated much more effectively than psychophysical tests of light perception, and also responds effectively to the sum total of diffuse and scattered light falling upon the entire retina. Therefore, the pupil light reflex may be an effective means of quantifying the transmission of light through the ocular wall by assessing the effect of pupil size (or lack thereof) on the amplitude of the pupil light reflex, as was demonstrated in this study. This may provide a means of assessing a patient's risk of light-associated damage to the eye. 

There are a number of potentially important implications resulting from this study. First, it predicts that in less pigmented patients, the presence of anisocoria would have much less effect on the estimation of the RAPD. This could become important in situations, such as acute trauma, where the presence and log unit amount of an RAPD may influence clinical decisions about the need for further evaluation and treatment. Ocular pigmentation, and hence penetration of light through the eye wall to the retina, would determine the effect of either miosis or mydriasis on the RAPD. In either situation, a lightly pigmented eye would minimize the effect of anisocoria and a darkly pigmented eye would maximize the effect. Second, in conditions, such as retinal degeneration or cataract development, which may be affected adversely by the total cumulative effect of a long-term exposure to ultraviolet and visible light, persons with less ocular pigmentation may be at greater risk for light toxicity, ^8–23 regardless of their pupil size in daylight. Finally, the results of our study suggested a physiologic explanation for differences in subjective light sensitivity and light scatter among dark-eyed and light-eyed persons. The effect of stray light entering the eye through an extrapupillary pathway (transmission through iris/uvea) was found to be significant in lightly pigmented subjects, in terms of its effect on the pupillary light reflex regardless of pupil size. However, its effect on visual function was not assessed in our study. The afferent input of light in the pupillary light reflex pathway is summated spatially over a very large area of retina, due to the large receptive field of the melanopsin retinal ganglion cells and the summation of their inputs by the recipient neurons in the pretectal olivary nucleus. This makes it difficult to equate stray light effects on the pupil light reflex with effects on visual perception. Recently, a study of stray light effects on visual function in humans did not find an association between macular pigment, using the compensation comparison heterochromatic flicker method with a visual foveal target (green target, 1° in size) flickering in counter-phase with a small stray light source (1.75° red target located at 8° in the periphery) using subjective visual perception. ³⁸ In this study, iris color also did not appear to influence significantly the effect of a small focal source of stray light on visual perception. It is possible that a more diffuse and brighter source of stray light that penetrates the iris, similar to what was used in our study, could have much greater effects on visual perception of small targets, and could be influenced more significantly by iris and uveal melanin pigmentation. Future studies should continue to take into consideration the role of ocular pigmentation on retinal illumination in terms of the effects of extrapupillary light transmission on visual perception, glare, and phototoxicity. 

Acknowledgments

Supported by an unrestricted grant from Research to Prevent Blindness (New York, New York), and the Division of Rehabilitation, Research and Development (Center of Excellence Grant) from the Veterans Administration, Washington, DC (RHK), and a visiting scholarship from Kyungpook National University Hospital, Taegu, Korea (SH). 

Disclosure: R.H. Kardon, None; S. Hong, None; A. Kawasaki, None