Pupillary Responses to High-Irradiance BlueLight Correlate with Glaucoma Severity

Annadata V. Rukmini, MBBS, MD,1 Dan Milea, MD, PhD,1,2 Mani Baskaran, DO, DNB,2,3

Alicia C. How, FRCS,2 Shamira A. Perera, FRCOphth,2 Tin Aung, PhD, FRCS,2,3 Joshua J. Gooley, PhD1,4

Purpose: To evaluate whether a chromatic pupillometry test can be used to detect impaired function ofintrinsically photosensitive retinal ganglion cells (ipRGCs) in patients with primary open-angle glaucoma (POAG)and to determine if pupillary responses correlate with optic nerve damage and visual loss.

Design: Cross-sectional study.Participants: One hundred sixty-one healthy controls recruited from a community polyclinic (55 men; 151

ethnic Chinese) and 40 POAG patients recruited from a glaucoma clinic (22 men; 35 ethnic Chinese) 50 years ofage or older.

Methods: Subjects underwent monocular exposure to narrowband blue light (469 nm) or red light (631 nm)using a modified Ganzfeld dome. Each light stimulus was increased gradually over 2 minutes to activatesequentially the rods, cones, and ipRGCs that mediate the pupillary light reflex. Pupil diameter was recordedusing an infrared pupillography system.

Main Outcome Measures: Pupillary responses to blue light and red light were compared between controlsubjects and those with POAG by constructing dose-response curves across a wide range of corneal irradiances(7e14 log photons/cm2 per second). In patients with POAG, pupillary responses were evaluated relative tostandard automated perimetry testing (Humphrey Visual Field [HVF]; Carl Zeiss Meditec, Dublin, CA) and scan-ning laser ophthalmoscopy parameters (Heidelberg Retinal Tomography [HRT]; Heidelberg Engineering, Heidel-berg, Germany).

Results: The pupillary light reflex was reduced in patients with POAG only at higher irradiance levels, cor-responding to the range of activation of ipRGCs. Pupillary responses to high-irradiance blue light associated morestrongly with disease severity compared with responses to red light, with a significant linear correlation observedbetween pupil diameter and HVF mean deviation (r ¼ �0.44; P ¼ 0.005) as well as HRT linear cup-to-disc ratio(r ¼ 0.61; P < 0.001) and several other optic nerve head parameters.

Conclusions: In glaucomatous eyes, reduced pupillary responses to high-irradiance blue light were associatedwith greater visual field loss and optic disc cupping. In POAG, a short chromatic pupillometry test that evaluates thefunction of ipRGCs can be used to estimate the degree of damage to retinal ganglion cells that mediate image-forming vision. This approach could prove useful in detecting glaucoma. Ophthalmology 2015;122:1777-1785 ª 2015 by the American Academy of Ophthalmology. This is an open access article under the CC BY-NC-NDlicense (http://creativecommons.org/licenses/by-nc-nd/4.0/).

The pupillary light reflex (PLR) is often used to assess theintegrity of the visual system. Until recently, however, thephotoreceptor pathways that drive pupillary light responseswere not well characterized. The afferent limb of the PLR isthought to be mediated solely by retinal ganglion cells(RGCs) that contain the short wavelength-sensitive photo-pigment melanopsin.1,2 Although melanopsin-containingRGCs are intrinsically photosensitive, they are also acti-vated extrinsically by rod and cone photoreceptors locatedin the outer retina.3e5 Genetic ablation of melanopsin RGCsin mice eliminates pupillary responses to light, indicatingthat these cells serve as a necessary conduit for light in-formation to reach the olivary pretectal nucleus in themidbrain.6

Growing evidence indicates that measuring pupillaryresponses to different wavelengths and irradiances of light(i.e., chromatic pupillometry) can be used to assess inner

� 2015 by the American Academy of OphthalmologyThis is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/). Published by Elsevier Inc.

versus outer retinal degeneration, because intrinsicallyphotosensitive RGCs (ipRGCs) and visual photoreceptorsdiffer in their response properties. In the absence ofrodecone input, melanopsin cells are preferentially sensitiveto blue light (lmax, approximately 480 nm), respond slug-gishly to light onset and offset, and are less sensitive to lightthan rods and cones.3,4,7e9 By comparison, rods are mostsensitive to bluish-green light (lmax, approximately 500nm), the photopic visual system is most sensitive to greenlight (lmax, 555 nm), and rodecone photoreceptors arecapable of driving fast pupillary responses. The wavelengthand irradiance of light exposure therefore can be manipu-lated to stimulate preferentially rods, cones, or the intrinsicmelanopsin response, thus providing a window onto thestatus of each photoreceptor cell type. For example, blue-light stimuli can be used to activate rods preferentially atlow irradiances and melanopsin at high irradiances, whereas

1777http://dx.doi.org/10.1016/j.ophtha.2015.06.002ISSN 0161-6420/15

Ophthalmology Volume 122, Number 9, September 2015

red-light stimuli can be used to target preferentially theactivation of middle- and long-wavelengthesensitivecones.10,11 As such, chromatic pupillometry using blue-lightand red-light stimuli could be used to detect photoreceptordysfunction associated with different types of retinaldiseases.

Recent studies suggest that melanopsin-containing RGCsare damaged in glaucoma, similar to nonmelanopsin RGCsthat mediate image-forming vision.12e19 Glaucoma is amajor cause of blindness, and hence early detection andtreatment are important for slowing the progression of thedisease. Patients often seek treatment late in the disease,however, because it is asymptomatic initially. The pupillarylight reflex is impaired in severe glaucoma,14,16 but there islimited evidence regarding whether reduced pupillary lightresponses correlate with glaucoma severity, as measured byvisual field (VF) testing and anatomic correlates of opticnerve damage. To address this limitation, we developed andtested a pupillometry-based protocol for evaluating photo-receptor dysfunction in which the light stimulus (blue orred) is increased gradually over time to assess sustainedpupillary responses across a wide range of irradiances. Theaim of our study was to determine if this chromaticpupillometry test can be used to detect loss of function ofipRGCs in patients with primary open-angle glaucoma(POAG) and if the degree of impairment in the PLRcorrelates with disease severity.

Methods

Subjects

Three hundred subjects 50 years of age or older were recruited toundergo a chromatic pupillometry test. Subjects were recruitedover a 2-month period (AugusteSeptember 2013) from a largerstudy involving more than 2000 volunteers who underwent astandardized ophthalmic examination at a community polyclinic.At the time of enrollment, subjects were either visiting the poly-clinic for minor health issues (nonocular) or accompanying anotherpatient at the clinic. The aim of the larger study was to evaluate theincidence of ocular abnormalities in older patients seeking outpa-tient medical care. The eye examination consisted of tests for visualacuity, intraocular pressure measurement by Goldmann applana-tion tonometry, automated refraction to assess refractive error, aslit-lamp examination, iris and fundus photography, and an ex-amination by a study ophthalmologist. The chromatic pupillometrytest was performed before examination by the ophthalmologist, andhence the researcher performing the pupillary recording was notaware of the ophthalmic status of subjects at the time of testing.

Forty-eight individuals diagnosed with POAG were recruitedover a 5-month period (October 2013eFebruary 2014) fromglaucoma clinics at the Singapore National Eye Centre. Patientswith POAG were defined by the following criteria: the presence ofglaucomatous optic neuropathy (defined as a loss of neuroretinalrim with a vertical cup-to-disc ratio of >0.7 or an intereye asym-metry of >0.2, notching attributable to glaucoma, or both) withcompatible VF loss (defined below), open angles on gonioscopy,and absence of secondary causes of glaucomatous opticneuropathy. Intraocular pressure before the start of glaucomatreatment was assessed by Goldmann applanation tonometry. Inaddition to the ophthalmic tests described for the polyclinic study,the glaucoma patients underwent standard automated perimetry

1778

(Humphrey Visual Field [HVF] Analyzer II model 750; Carl ZeissMeditec, Dublin, CA) and Heidelberg Retinal Tomography (HRT3; Heidelberg Engineering, Heidelberg, Germany), performedeither on the day of chromatic pupillometry testing or within thepreceding 3 months. Humphrey Visual Field testing was performedwith near refractive correction using the 24-2 Swedish interactivethresholding algorithm with stimulus size III. Repeat testing wasperformed if false-positive or false-negative responses exceeded33% or if the fixation loss rate was more than 20%. Patients whocould not achieve these reliability criteria were ineligible for thestudy. A glaucomatous VF defect was defined by the presence of aglaucoma hemifield test result outside normal limits and the pres-ence of at least 3 contiguous, nonedge test points within the samehemifield on the pattern deviation probability plot at P < 0.05, withat least 1 point at P < 0.01, excluding points directly above andbelow the blind spot. Glaucoma severity was graded according tothe Hodapp-Parrish-Anderson scale,20 in which subjects with meandeviation less than �6 dB were classified as having early VF loss,mean deviation between �6 dB and �12 dB as moderate VF loss,and mean deviation more than (i.e., more negative than) �12 dB assevere VF loss. For HRT, global and segmental disc and cup areaswere analyzed using the standard reference plane. In most cases,the POAG disease severity was known at the time that thepupillometry test was performed.

Control subjects and POAG patients were excluded fromchromatic pupillometry testing if they had undergone previousintraocular surgery. Additional exclusionary criteria for individualswith POAG included significant nuclear sclerosis of more thangrade 2 severity on slit-lamp examination; severe retinal or ocularcomorbid conditions including, but not limited to, diabetic reti-nopathy and age-related macular degeneration; and clinically sig-nificant pupillary abnormalities (except for relative afferentpupillary defects). Because the ophthalmic health of subjectsrecruited from the polyclinic was determined after chromaticpupillometry testing, data from participants who failed to meet theabove criteria were excluded post hoc from the analyses. There-fore, the number of subjects with normal ocular health was notdetermined a priori, whereas the sample size of the group withPOAG was determined before study recruitment. Demographicinformation was collected using interviewer-administered ques-tionnaires. The study was approved by the SingHealth CentralizedInstitutional Review Board, and all participants provided writteninformed consent. Research procedures adhered to ethical princi-ples outlined in the Declaration of Helsinki.

Chromatic Pupillometry

Before each light exposure, participants were seated with their headposition fixed by a chinrest for at least 1 minute in a dark envi-ronment. To measure the direct PLR, light was administered to oneeye using a modified Ganzfeld dome (Labsphere, Inc, North Sut-ton, NH), with the other eye covered by a patch. Subjects wereexposed to a blue-light stimulus (469 nm) or red-light stimulus(631 nm), with the order of exposure randomized and counter-balanced. Narrow-bandwidth light was provided using light-emitting diodes (Nichia Corporation, Tokushima, Japan) thatwere controlled using a function generator (Keithley Instruments,Inc, Cleveland, OH). Current was applied to the light-emittingdiodes over a 2-minute period using a logarithmically increasingfunction in 14 000 steps. Hence, the light stimulus was perceivedas increasing gradually in intensity over time, ranging from 6.8 logphotons/cm2 per second to 13.8 log photons/cm2 per second at thelevel of the cornea. The maximum light level corresponded to12.27 lux and 28.30 mW/cm2 for the blue-light stimulus and 20.43lux and 21.67 mW/cm2 for the red-light stimulus. Light levels werecalibrated using a portable radiometer (ILT1700 radiometer;

Rukmini et al � Pupillary Light Responses in Glaucoma

International Light Technologies, Peabody, MA), with the sensorplaced at the level of the subject’s eyes during light exposure. Thelight exposure was followed by a 1-minute period of darkness,during which the subject remained seated with his or her head inthe dome. This was followed immediately by the next lightexposure sequence for the other color of light (1 minute of dark-ness, 2 minutes of light, and 1 minute of darkness). In the eventthat a quality eye image could not be obtained for one of the trials,participants underwent a third light exposure sequence (7 of 348subjects). During each 4-minute light exposure sequence, subjectswore a head-mounted eye tracking device that recorded monocularpupil diameter at a rate of 120 samples per second (ETL-100HPupillometry Lab; ISCAN, Inc, Woburn, MA). The eye tracker wasconfigured to measure pupillary responses of the left eye in allstudy participants, with the exception of 5 glaucoma patients whohad detectable disease only in the right eye.

Data Analysis and Statistics

Pupillometry recordings from all participants were coded andprocessed without knowledge of the patient’s ocular health statusor POAG disease severity. For each light exposure, pupil diametermeasurements were processed to remove blink artifacts in therecording, and then expressed as a percentage of the median pupildiameter during the preceding dark period. Data were reduced bytaking the median pupillary constriction response in 0.5 log unitbins from 7 to 14 log photons/cm2 per second, resulting in 14 datapoints per light exposure sequence. The interaction of irradiancewith wavelength (blue vs. red) or with ocular health (controls vs.glaucomatous eyes) on pupillary responses was assessed using a 2-way repeated-measures analysis of variance. For those compari-sons in which the omnibus test reached statistical significance,pairwise multiple comparison procedures were performed using theHolm-Sidak method. Differences in age between controls and pa-tients with POAG were assessed using an unpaired Student’s t testassuming unequal variance. Within the POAG patient group, thestrength of the linear relationship between pupillary constrictionand clinical measures (e.g., HVF mean deviation and linear opticcup-to-disc ratio) was assessed using Pearson’s correlation anal-ysis. For all statistical tests, the threshold for significance was set ata ¼ 0.05. Statistics were performed using Sigmaplot softwareversion 12.0 (Systat Software, Inc, San Jose, CA) and SPSS soft-ware version 22 (IBM Corp, Armonk, NY).

Results

Subject Characteristics

Of the 300 subjects who were studied in the polyclinic, 86 in-dividuals were excluded from the analysis because of diagnosis ofan ophthalmic condition (e.g., severe cataracts, primary angle-closure glaucoma, diabetic retinopathy) and were referred to anophthalmologist for a follow-up assessment. In the remaining 214

Table 1. Subject C

Subject Group No. No. of Men No

Controls 161 55No cataract 54 36Mild cataract 107 19

Glaucoma (stage) 40 22Early 19 9Moderate 10 4Severe 11 9

subjects with normal-for-age ocular health, we excluded data from43 participants based on the criterion that the baseline pupildiameter measured in darkness differed between light exposuretrials by more than 10%. Because pupillary light responses weremeasured relative to baseline, the 10% cutoff ensured that themagnitude of pupillary constriction could be compared reliablybetween blue-light and red-light exposures and across groups withnormal ocular health and POAG. Technical problems resulted inloss of data in an additional 10 individuals. Therefore, a total of161 subjects were included in the control group, among whom 107individuals had mild cataracts. Pupillary responses in these subjectswere similar to those observed in individuals without cataracts(F1,159 < 2.30; P> 0.13, for main effect of group for both colors oflight). Of the 48 POAG patients recruited from the glaucoma clinic,baseline pupil diameter was unstable in 7 individuals, and data lossoccurred during the pupillometry recording in 1 participant. Wetherefore included data from 40 patients in our analyses,comprising persons with early (n ¼ 19), moderate (n ¼ 10), andsevere (n ¼ 11) POAG. Patients with POAG were 4 years older onaverage (t ¼ 3.6; P < 0.001) relative to controls and included agreater proportion of men and ethnic Chinese individuals (Table 1).

Chromatic Pupillometry in Control Subjects

In response to the gradually increasing light stimulus (Fig 1A), theonset of pupillary constriction in control subjects occurred muchearlier for blue light relative to red light (Fig 1B, C). Pupildiameter closely tracked the change in irradiance over time forboth exposures, followed by a similar time course of redilationafter light offset. Pupillary light responses measured over timethen were converted to dose-response curves by grouping resultsin 0.5-log unit bins based on photon density (Fig 1D). Thethreshold for pupillary constriction was between 9.0 and 9.5 logphotons/cm2 per second in response to blue light, as comparedwith 10.5 to 11.0 log photons/cm2 per second for exposure tored light. Pupillary constriction responses were significantlygreater in response to blue light across a wide range ofirradiances, from 9.0 to 13.0 log photons/cm2 per second(F13,2080 ¼ 62.18; P < 0.001 for interaction; t > 3.35 and P <0.05 for all pairwise comparisons), with the greatest difference inspectral responses observed in the middle of this range (Fig 1E).

Chromatic Pupillometry in Patients with Glaucoma

Similar to control subjects, the onset of pupillary constriction inpatients with POAG occurred earlier for the blue-light stimulus.This was evident across different disease stages of POAG, rangingfrom early to severe (Fig 2AeC). As with controls, glaucomatouseyes showed stronger pupillary responses to blue light in theirradiance range from 9.0 to 13.0 log photons/cm2 per second(F13,505 ¼ 9.89; P < 0.001 for interaction; t > 2.02 and P <0.05 for all pairwise comparisons; Fig 2D), and the greatestdifference in pupillary constriction relative to red light was in the

haracteristics

. of Chinese Individuals Age (Mean ± Standard Deviation)

151 59.8�6.249 56.4�5.1102 61.5�6.035 63.8�6.116 62.3�6.310 64.9�6.19 65.5�5.6

1779

Figure 1. Graphs showing chromatic pupillometry results in control pa-tients with normal-for-age ocular health. A, Patients were exposed to a4-minute light exposure sequence consisting of 1 minute of darkness,2 minutes of monocular exposure to a gradually increasing blue-light (469nm) or red-light (631 nm) stimulus, and 1 minute of darkness after lightoffset. B, Representative pupillary constriction responses are shown for aparticipant who underwent the pupillometry test. C, Average pupillaryresponse, expressed as a percentage of the dark pupil diameter, across 161control subjects. D, Dose-response curves for pupillary constriction duringexposure to blue light versus red light, with data grouped in 0.5-log unitbins. Asterisks indicate significant differences in percentage of pupillaryconstriction between light conditions. E, Mean difference in the pupillaryconstriction response for blue light versus red light, demonstrating greaterresponses to blue light across a 4-log unit range, from 9.0 to 13.0 logphotons/cm2 per second. In (C), (D), and (E), the mean � standard errorof the mean is shown.

Figure 2. Graphs showing chromatic pupillometry results in patients withprimary open-angle glaucoma (POAG). Representative pupillaryconstriction responses are shown for patients with (A) early, (B) moderate,or (C) severe glaucoma. The pupillary constriction response was reduced inpatients with greater disease severity. As in Figure 1, patients were exposedto 1 minute of darkness, 2 minutes of monocular exposure to a graduallyincreasing blue-light (469 nm) or red-light (631 nm) stimulus, and 1minute of darkness. D, Dose-response curves for pupillary constrictionduring exposure to blue light versus red light, assessed in 40 patients withPOAG. Asterisks indicate significant differences in the response betweenlight conditions. E, Patients exhibited greater responses to blue light in therange from 9.0 to 13.0 log photons/cm2 per second, based on the meandifference in pupillary constriction for blue light versus red light. In(D) and (E), the mean � standard error of the mean is shown.

Ophthalmology Volume 122, Number 9, September 2015

middle of this irradiance range (Fig 2E). Overall, dose-responsecurves to blue light and red light were similar in shape betweencontrols and patients with POAG; hence, the wavelength de-pendency of pupillary responses seemed to be preserved in glau-comatous eyes.

Deficits in Pupillary Responses to Light in PrimaryOpen-Angle Glaucoma at Higher Irradiances

Next, we evaluated whether the magnitude of pupillary light re-sponses differed in controls and patients with POAG. At lowerirradiances, there was no difference between groups in pupillaryconstriction for blue-light and red-light stimuli (Fig 3A, B). As thecorneal irradiance was increased beyond 11.5 log photons/cm2 persecond, however, the relative response in patients with POAGbecame increasingly impaired. For both colors of light, there wasa significant interaction between group (POAG vs. controls) andirradiance such that the difference in pupillary constrictionbetween healthy and glaucomatous eyes was greatest at thehighest irradiances tested (F13,2587 > 11.38; P < 0.001 for blue-and red-light stimuli; t > 2.73 and P < 0.05 for all pairwisecomparisons of more than 11.5 log photons/cm2 per second). To

1780

address the possibility that the difference in pupillary responsesbetween groups may be driven by small differences in age or otherpatient characteristics, a secondary analysis was performed inwhich 40 individuals were selected randomly from the controlgroup who were matched for age (�2 years), sex, ethnicity, andorder of light exposure with patients who had POAG. Consistentwith our previous analysis, pupillary constriction was impaired athigher irradiances in glaucomatous eyes when these patients werematched with similar control participants (F13,1014 > 5.42; P <0.001; t > 2.28 and P < 0.05 for all pairwise comparisons of morethan 11.5 log photons/cm2 per second for 469-nm light and forcomparisons of more than 12.0 log photons/cm2 per second for631-nm light).

To explore whether pupillary responses differed for controlsversus POAG patients with early, moderate, or severe disease, ananalysis of variance was performed using the maximum pupillaryconstriction values for blue-light and red-light stimuli. The PLRdiffered significantly across groups during exposure to blue light(F3,197 ¼ 13.72; P < 0.001), such that pupillary constriction wasgreater for control subjects versus POAG patients with moderate orsevere VF loss (P < 0.025 for both comparisons, Fisher’s leastsignificant difference [LSD]), whereas the difference between

Figure 3. Graphs showing impaired pupillary constriction responses inpatients with primary open-angle glaucoma. Dose-response curves for pu-pillary constriction for controls (n ¼ 161, black traces) and patients withglaucoma (n ¼ 40) who were exposed to (A) blue 469-nm light (bluetrace), and (B) red 631-nm light (red trace). For both colors of light, themagnitude of the pupillary light reflex was reduced in glaucomatous eyes asthe irradiance of light was increased (>11.5 log photons/cm2 per second).Pupil diameter is expressed as a percentage of the dark pupil measuredbefore each light exposure. Asterisks show significant differences in pu-pillary responses between controls and patients with glaucoma. The mean� standard error of the mean is shown.

Rukmini et al � Pupillary Light Responses in Glaucoma

control subjects and individuals with early glaucoma failed to reachstatistical significance by the smallest of possible margins (P ¼0.050). Pupillary constriction also differed across groups duringexposure to red light (F3,197 ¼ 10.02; P < 0.001), such that re-sponses in control subjects were greater than those observed forPOAG patients with early, moderate, or advanced VF loss (P <0.045 for each paired comparison, Fisher LSD). It should be noted,however, that the sample size was small across different diseaseseverity groups, and the experiment was not designed to have abalanced number of participants across groups.

Correlation between Pupillary Responses inGlaucoma and Clinical Measures

Based on deficits in the PLR observed in patients with POAG, wetested whether the pupillary constriction response to high-irradianceblue light (>13.5 log photons/cm2 per second) correlated withclinical measures used for assessing severity of glaucoma.21 Pupil

Figure 4. Scatterplots showing the correlation between pupillary responsesand glaucoma severity. In patients with primary open-angle glaucoma, thepupillary constriction response to high-irradiance blue light (>13.5 logphotons/cm2 per second) correlated significantly with visual field lossassessed by (A) Humphrey Visual Field (HVF) testing, as well as (B) thelinear cup-to-disc ratio determined by Heidelberg Retinal Tomography.The linear regression line is shown with 95% confidence intervals. dB ¼decibels.

diameter (expressed as a percentage of the dark pupil) exhibited asignificant correlation with HVF mean deviation (Fig 4A;r ¼ �0.44; P ¼ 0.005) and linear cup-to-disc ratio assessed byHRT (Fig 4B; r ¼ 0.61; P � 0.001). That is, impaired pupillaryresponses were associated with greater VF loss and cupping of theoptic disc. Notably, the correlation between the PLR and thelinear cup-to-disc ratio was just as strong as the correlation be-tween HVF mean deviation and the linear cup-to-disc ratio(r¼�0.59; P� 0.001). By comparison, intraocular pressure beforetreatment for glaucoma did not correlate significantly with pupillaryconstriction, HVF mean deviation, or the linear cup-to-disc ratio(jrj < 0.14 and P > 0.40 for all comparisons). Also, the patternstandard deviation obtained via HVF testing did not correlate withpupillary responses to blue light or red light (jrj< 0.31 and P> 0.49for both comparisons). Next, we systematically evaluated the rela-tionship between the magnitude of pupillary constriction at differentirradiances with HVF mean deviation and optic nerve head pa-rameters assessed by HRT. As the irradiance of light was increased,the strength of the correlation between pupillary constriction andHVF and HRT measures also increased. For blue-light and red-lightstimuli, a moderate correlation was observed for the PLR and HVFmean deviation by the end of the light exposure (Fig 5). Bycomparison, the relationship between pupillary responses andHRT parameters was much stronger for blue light relative to redlight, with correlation coefficients reaching their highest levelsnear the end of the blue-light stimulus for several measuresincluding cup area (r ¼ 0.60; P < 0.001), cup volume (r ¼ 0.56;P < 0.001), cup-to-disc area ratio (r ¼ 0.69; P < 0.001), and meanretinal nerve fiber layer thickness (r ¼ �0.47; P ¼ 0.022).

Discussion

Our results suggest that the functional integrity of ipRGCsand their efferent projections can be evaluated by measuringpupillary responses to a light stimulus that is increasedgradually over time. Using this approach, it was possible toconstruct dose-response curves to light over a short timeinterval. Because melanopsin-dependent pupillary responseshave a relatively high threshold of activation and are short-wavelength sensitive, the ramp-up stimulus that we usedallows for examination of rodecone and melanopsin-dependent RGC activation using a single light-exposuresequence. In glaucomatous eyes, deficits in the PLR wereobserved primarily at higher irradiances, and responses toblue light correlated most strongly with VF loss severity andcupping of the optic disc. These findings suggest that inPOAG, the magnitude of impairment of melanopsin-dependent pupillary responses potentially can be used toestimate the degree of damage to RGCs that mediate image-forming vision.

Although it has been suggested that ipRGCs may be sparedpreferentially in animal models with experimentally inducedocular hypertension,22 we and other investigators have foundthat pupillary light responses are impaired in patients withglaucoma.14,16,17 Here, deficits in pupillary responses to bluelight were specific to the irradiance range associated withactivation of melanopsin-containing RGCs.7,8,23 Because rod-driven pupillary light responses are thought to be mediatedsolely through ipRGCs,6,24 it may be expected that loss ofipRGC function also would result in reduced responses to low-irradiance light. Under the lighting conditions used in thepresent study, however, the amount of rod input to ipRGCs in

1781

Blue Light (469nm)

8 9 10 11 12 137 14 8 9 10 11 12 137 14

HVF Mean Deviation (dB)

Cup Area (mm2)

Cup Volume (mm3)

Mean RNFL Thickness (mm)

Cup Shape Measure

Rim Volume (mm3)

Mean Cup Depth (mm)

RNFL Cross-sectional Area (mm2)

Disc Area (mm2)

Rim Area (mm2)

Height Variation Contour (mm)

Maximum Cup Depth (mm)0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Red Light (631nm)

Linear Cup-to-disc Ratio

Global Rim-to-disc Area Ratio

Cup-to-disc Area Ratio

Temporal Rim-to-disc Area Ratio

Temporal Inferior Rim-to-disc Area Ratio

Temporal Superior Rim-to-disc Area Ratio

Nasal Rim-to-disc Area Ratio

Nasal Superior Rim-to-disc Area Ratio

Nasal Inferior Rim-to-disc Area Ratio

Log Photons/cm2 s1 Log Photons/cm2 s1

Figure 5. Pupillary constriction responses correlated with clinical measures used to diagnose glaucoma. The heat maps show Pearson’s correlation coefficient(absolute values) for visual field testing and optic nerve head parameters versus pupillary constriction in a group of 40 patients with primary open-angleglaucoma. Correlations with pupillary responses to blue light (469 nm) and red light (631 nm) are shown in 0.5-log unit bins from 7 to 14 log pho-tons/cm2 per second. Warmer colors (i.e., more red) indicate higher correlation coefficient values. Results for Humphrey Visual Field (HVF) analysis andHeidelberg Retinal Tomography correlated most strongly with the magnitude of pupillary constriction during exposure to high-irradiance blue light.dB ¼ decibels; RNFL ¼ retinal nerve fiber layer.

Ophthalmology Volume 122, Number 9, September 2015

POAGpatients seemed tobe sufficient todrive a normalPLR inthe scotopic visual range. Therefore, deficits in ipRGC functionmeasured by pupillometry may not relate directly to loss ofperipheral vision that occurs in glaucoma. Our findings couldbe explained in part by regional differences in cell density,synaptic contacts, and receptive fields of melanopsin cellsversus RGCs that mediate vision.4 Pupillary responses to redlight in the photopic visual range suggest involvement ofcone photoreceptors,8 but also could be mediated by rods.24

Notably, patients with POAG showed normal spectralresponses to light, i.e., the difference in pupillary constrictionfor blue light versus red light was similar to that of controls,suggesting that the relative contribution of melanopsin androdecone photoreceptors to the PLR was not substantiallyaffected by the disease.

Growing evidence indicates that chromatic pupillometry-based approaches can be used to detect abnormalities that

1782

associate with damage either to the photoreceptor layer or toRGCs in different types of retinal diseases.10 In this study,pupillary responses were examined across much finersteps in irradiance compared with prior work,25e27 whichmade it possible to determine the dose response across awide range of irradiance levels. In principle, this approachshould perform better at detecting subtle deficits in the PLR,hence providing more detailed information regarding theorigin and magnitude of the underlying neuroanatomic ab-normality. Light stimuli also were equated by photon den-sity, which is better for making inferences regarding thecontribution of different photoreceptor types to the lightresponse.28

Several studies have used the postillumination pupillaryresponse (PIPR) to evaluate melanopsin-dependent RGCresponses.14,16,29 Because the ipRGCs respond sluggishly tochanges in lighting, the PIPR is thought to reflect sustained

Rukmini et al � Pupillary Light Responses in Glaucoma

activation of the melanopsin pathway after light offset.7,30 Inpatients with advanced glaucoma, the PIPR to high-irradiance blue light is blunted relative to the response innormal subjects, and the difference in the PIPR betweenblue-light and red-light stimuli correlates inversely withHVF mean deviation in patients with glaucomatouseyes.14,16 Glaucoma also can be detected based on a relativeafferent pupillary defect (RAPD), in which asymmetricpupillary responses are observed because of a difference indisease progression between eyes.31,32 Automated RAPDpupillometry systems have been used to screen for patientswith glaucoma who exhibit asymmetric disease33e35; how-ever, complementary approaches are still needed becausemany patients show bilateral damage to the optic nervewithout an apparent RAPD. Although the aforementionedpupillometry-based methods show promise for detectingimpaired photic transmission by ipRGCs, further improve-ments are needed if they are to be used to screen reliably forglaucoma in population studies. Similarly, our light expo-sure protocol likely requires further optimization if it is to beused for distinguishing early stages of glaucoma fromnormal ocular health. It is possible that combining ourapproach with other methods (e.g., measuring the PIPR andRAPD) could be used to improve detection of glaucoma,which remains to be tested.

A limitation of our study is that light stimuli were cali-brated at the level of the cornea, rather than attempting tocorrect for prereceptoral filtering by the cornea and lens. Inthe age group studied here (50 years and older), lens yel-lowing and mild cataracts are common, which would beexpected to reduce the amount of short-wavelength light thatreaches the retina.36 In addition, because pupillaryconstriction was short-wavelength sensitive and measure-ments were performed for the same eye that was beingexposed to light, the smaller pupil in the blue-light conditionfurther limited the number of blue photons reaching theretina. Therefore, our method likely underestimated therelative sensitivity of pupillary responses to blue light versusred light at the level of retinal photoreceptors. Given that theprimary deficit we observed in POAG patients was areduction in the pupillary response at high irradiances, asingle blue-light exposure sequence may be sufficient if ourapproach is to be adapted for use as a screening test. Thered-light stimulus nonetheless may be useful for detectingdiseases that affect cone function.26

Although previous studies suggest that pupillary re-sponses to blue-light and red-light stimuli are reproducibleacross different times of day in healthy subjects (intraclasscorrelation coefficient, �0.7 for maximum pupillaryconstriction),37 testeretest reliability was not assessed in thepresent study. Also, subjects in the control group did notundergo VF testing or HRT; hence, these participants werenot included in analyses on the relationship betweenpupillary light responses with visual loss and cupping ofthe optic disc. It also should be highlighted that deficits inthe PLR observed in POAG patients may be similar forother conditions affecting ipRGC transmission. Forexample, it was shown recently that nonarteritic anteriorischemic optic neuropathy is associated with an abnormalPIPR to blue light, which is consistent with dysfunction of

ipRGCs.38 Hence, a reduction in pupillary responses tohigh-irradiance blue light does not necessarily imply that apatient has glaucoma, but rather suggests reduced ipRGCactivation or optic nerve transmission. In future work, it willbe important to characterize pupillary responses further usingour testing protocol in other patient populations with damageto the optic nerve, including other types of glaucoma. It willalso be important to determine whether our results aregeneralizable to patient groups that differ in ethnicity andage, because our subject pool included predominantly ethnicChinese individuals over a limited age range.

Our study provides proof of concept that a short-durationchromatic pupillometry test can be used to assess loss ofipRGC function associated with POAG. Because gradualloss of peripheral vision in glaucomatous eyes can go un-noticed for many years, patients often seek treatment aftersubstantial and irreversible damage to the optic nerve hasoccurred. We and others therefore have explored chromaticpupillometry as an approach for detecting early retinaldysfunction. A major advantage of pupillometry-basedmethods is that they can be adapted for populationscreening in the form of portable devices, for example, as adesktop system or incorporated into specialized gog-gles,34,37,39 that can provide objective data on the integrityof the pathway from the retina to the midbrain. If pupill-ometry systems are shown to provide reliable detection ofocular diseases using short testing protocols, such low-costand patient-friendly technologies could be used to screenfor damage to the visual system in population studies andcommunity samples. This would allow for identification ofat-risk individuals who should undergo a comprehensiveophthalmic examination to treat or halt the progression ofPOAG or other ocular diseases.

References

1. Hattar S, Lucas RJ, Mrosovsky N, et al. Melanopsin and rod-cone photoreceptive systems account for all major accessoryvisual functions in mice. Nature 2003;424:76–81.

2. Panda S, Provencio I, Tu DC, et al. Melanopsin is required fornon-image-forming photic responses in blind mice. Science2003;301:525–7.

3. Berson DM, Dunn FA, Takao M. Phototransduction by retinalganglion cells that set the circadian clock. Science 2002;295:1070–3.

4. Dacey DM, Liao HW, Peterson BB, et al. Melanopsin-expressing ganglion cells in primate retina signal colour andirradiance and project to the LGN. Nature 2005;433:749–54.

5. Wong KY, Dunn FA, Graham DM, Berson DM. Synapticinfluences on rat ganglion-cell photoreceptors. J Physiol2007;582(Pt 1):279–96.

6. Guler AD, Ecker JL, Lall GS, et al. Melanopsin cells are theprincipal conduits for rod-cone input to non-image-formingvision. Nature 2008;453:102–5.

7. Gamlin PD, McDougal DH, Pokorny J, et al. Human andmacaque pupil responses driven by melanopsin-containingretinal ganglion cells. Vision Res 2007;47:946–54.

8. Gooley JJ, Ho Mien I, St Hilaire MA, et al. Melanopsin androd-cone photoreceptors play different roles in mediating pu-pillary light responses during exposure to continuous light inhumans. J Neurosci 2012;32:14242–53.

1783

Ophthalmology Volume 122, Number 9, September 2015

9. Zaidi FH, Hull JT, Peirson SN, et al. Short-wavelength lightsensitivity of circadian, pupillary, and visual awareness inhumans lacking an outer retina. Curr Biol 2007;17:2122–8.

10. Feigl B, Zele AJ. Melanopsin-expressing intrinsically photo-sensitive retinal ganglion cells in retinal disease. Optom VisSci 2014;91:894–903.

11. La Morgia C, Ross-Cisneros FN, Hannibal J, et al. Melanop-sin-expressing retinal ganglion cells: implications for humandiseases. Vision Res 2011;51:296–302.

12. de Zavalia N, Plano SA, Fernandez DC, et al. Effect ofexperimental glaucoma on the non-image forming visual sys-tem. J Neurochem 2011;117:904–14.

13. Drouyer E, Dkhissi-Benyahya O, Chiquet C, et al. Glaucomaalters the circadian timing system. PLoS One 2008;3:e3931.

14. Feigl B, Mattes D, Thomas R, Zele AJ. Intrinsically photo-sensitive (melanopsin) retinal ganglion cell function in glau-coma. Invest Ophthalmol Vis Sci 2011;52:4362–7.

15. Jakobs TC, Libby RT, Ben Y, et al. Retinal ganglion celldegeneration is topological but not cell type specific in DBA/2J mice. J Cell Biol 2005;171:313–25.

16. Kankipati L, Girkin CA, Gamlin PD. The post-illuminationpupil response is reduced in glaucoma patients. Invest Oph-thalmol Vis Sci 2011;52:2287–92.

17. Nissen C, Sander B, Milea D, et al. Monochromatic pupillo-metry in unilateral glaucoma discloses no adaptive changessubserved by the ipRGCs. Front Neurol 2014;5:15.

18. Wang HZ, Lu QJ, Wang NL, et al. Loss of melanopsin-containing retinal ganglion cells in a rat glaucoma model.Chin Med J (Engl) 2008;121:1015–9.

19. Zhang Q, Vuong H, Huang X, et al. Melanopsin-expressingretinal ganglion cell loss and behavioral analysis in the Thy1-CFP-DBA/2J mouse model of glaucoma. Sci China Life Sci2013;56:720–30.

20. Hodapp E, Parrish RK, Anderson DR. Clinical decisions inglaucoma. St. Louis: C.V. Mosby Company; 1993.

21. Zangwill LM, Weinreb RN, Beiser JA, et al. Baseline topo-graphic optic disc measurements are associated with thedevelopment of primary open-angle glaucoma: the ConfocalScanning Laser Ophthalmoscopy Ancillary Study to theOcular Hypertension Treatment Study. Arch Ophthalmol2005;123:1188–97.

22. Li RS, Chen BY, Tay DK, et al. Melanopsin-expressing retinalganglion cells are more injury-resistant in a chronic ocular hy-pertension model. Invest Ophthalmol Vis Sci 2006;47:2951–8.

23. Lucas RJ, Hattar S, Takao M, et al. Diminished pupillary lightreflex at high irradiances in melanopsin-knockout mice. Sci-ence 2003;299:245–7.

24. Altimus CM, Guler AD, Alam NM, et al. Rod photoreceptorsdrive circadian photoentrainment across a wide range of lightintensities. Nat Neurosci 2010;13:1107–12.

25. Kardon R, Anderson SC, Damarjian TG, et al. Chromatic pupilresponses: preferential activation of the melanopsin-mediated

1784

versus outer photoreceptor-mediated pupil light reflex.Ophthalmology 2009;116:1564–73.

26. Kardon R, Anderson SC, Damarjian TG, et al. Chromaticpupillometry in patients with retinitis pigmentosa. Ophthal-mology 2011;118:376–81.

27. Leon L, Crippa SV, Borruat FX, Kawasaki A. Differentialeffect of long versus short wavelength light exposure on pu-pillary re-dilation in patients with outer retinal disease. ClinExperiment Ophthalmol 2012;40:e16–24.

28. Lucas RJ, Peirson SN, Berson DM, et al. Measuring and usinglight in the melanopsin age. Trends Neurosci 2014;37:1–9.

29. Park JC, Moura AL, Raza AS, et al. Toward a clinical protocolfor assessing rod, cone, and melanopsin contributions to thehuman pupil response. Invest Ophthalmol Vis Sci 2011;52:6624–35.

30. Kankipati L, Girkin CA, Gamlin PD. Post-illumination pupilresponse in subjects without ocular disease. Invest OphthalmolVis Sci 2010;51:2764–9.

31. Chang DS, Boland MV, Arora KS, et al. Symmetry of thepupillary light reflex and its relationship to retinal nerve fiberlayer thickness and visual field defect. Invest Ophthalmol VisSci 2013;54:5596–601.

32. Chang DS, Xu L, Boland MV, Friedman DS. Accuracy ofpupil assessment for the detection of glaucoma: a systematicreview and meta-analysis. Ophthalmology 2013;120:2217–25.

33. Chang DS, Arora KS, Boland MV, et al. Development andvalidation of an associative model for the detection of glau-coma using pupillography. Am J Ophthalmol 2013;156:1285–1296e2.

34. Ozeki N, Yuki K, Shiba D, Tsubota K. Pupillographic eval-uation of relative afferent pupillary defect in glaucoma pa-tients. Br J Ophthalmol 2013;97:1538–42.

35. Tatham AJ, Meira-Freitas D, Weinreb RN, et al. Detectingglaucoma using automated pupillography. Ophthalmology2014;121:1185–93.

36. Kessel L, Lundeman JH, Herbst K, et al. Age-related changesin the transmission properties of the human lens and theirrelevance to circadian entrainment. J Cataract Refract Surg2010;36:308–12.

37. Herbst K, Sander B, Milea D, et al. Test-retest repeatability ofthe pupil light response to blue and red light stimuli in normalhuman eyes using a novel pupillometer. Front Neurol 2011;2:10.

38. Herbst K, Sander B, Lund-Andersen H, et al. Unilateralanterior ischemic optic neuropathy: chromatic pupillometry inaffected, fellow non-affected and healthy control eyes. FrontNeurol 2013;4:52.

39. Lorenz B, Strohmayr E, Zahn S, et al. Chromatic pupillometrydissects function of the three different light-sensitive retinalcell populations in RPE65 deficiency. Invest Ophthalmol VisSci 2012;53:5641–52.

Footnotes and Financial Disclosures

Originally received: November 28, 2014.Final revision: April 27, 2015.Accepted: June 2, 2015.Available online: August 20, 2015. Manuscript no. 2014-1910.1 Program in Neuroscience and Behavioral Disorders, DukeeNationalUniversity of Singapore Graduate Medical School, Singapore, Republic ofSingapore.2 Singapore Eye Research Institute, Singapore National Eye Center,Singapore, Republic of Singapore.

3 Department of Ophthalmology, Yong Loo Lin School of Medicine, Na-tional University of Singapore, Singapore, Republic of Singapore.4 Department of Physiology, Yong Loo Lin School of Medicine, NationalUniversity of Singapore, Singapore, Republic of Singapore.

Financial Disclosure(s):The author(s) have made the following disclosure(s):

D.M.: Patent d Chromatic pupillometry test procedures described herein(Singapore patent application no. 10201408408U)

Rukmini et al � Pupillary Light Responses in Glaucoma

J.J.G.: Patent e Chromatic pupillometry test procedures described herein(Singapore patent application no. 10201408408U)

Supported by the DukeeNational University of Singapore SignatureResearch Program funded by the Agency for Science, Technology andResearch, Singapore, and the Ministry of Health, Singapore, Republic ofSingapore; the National Medical Research Council, Singapore, Republic ofSingapore (grant nos.: NMRC/NIG/1000/2009 [J.J.G.] and NMRC/CIRG/1401/2014 [D.M.]); the Singapore National Eye Centre Health ResearchEndowment Fund, Singapore, Republic of Singapore (grant no.: 1005/20/2013 [T.A., A.C.H.]); and the Biomedical Research Council, Singapore,Republic of Singapore (grant nos.: 01-TCRP-2010 and TCR0101674[T.A.]). The funding organizations had no role in the design and conduct ofthe research.

Author Contributions:

Conception and design: Rukmini, Milea, Aung, Gooley

Analysis and interpretation: Rukmini, Milea, Baskaran, How, Perera, Aung,Gooley

Data collection: Rukmini, Baskaran, How, Perera

Obtained funding: Milea, How, Aung, Gooley

Overall responsibility: Rukmini, Milea, Baskaran, How, Perera, Aung,Gooley

Abbreviations and Acronyms:HRT ¼ Heidelberg Retinal Tomography; HVF ¼ Humphrey Visual Field;ipRGC ¼ intrinsically photosensitive retinal ganglion cell;PIPR ¼ postillumination pupillary response; PLR ¼ pupillary light reflex;POAG ¼ primary open-angle glaucoma; RGC ¼ retinal ganglion cell;RAPD ¼ relative afferent pupillary defect; VF ¼ visual field.

Correspondence:Joshua J. Gooley, PhD, Program in Neuroscience and Behavioral Disorders,DukeeNational University of Singapore Graduate Medical SchoolSingapore, 8 College Road Singapore 169857, Republic of Singapore. E-mail: joshua.gooley@duke-nus.edu.sg.

1785