- Email: [email protected]
Chromatic Multifocal Pupillometer for Objective Perimetry and Diagnosis of Patients with Retinitis Pigmentosa
Chromatic Multifocal Pupillometer for Objective Perimetry and Diagnosis of Patients with Retinitis Pigmentosa Ron Chibel, MSc,1,2 Ifat Sher, PhD,1,2 Daniel Ben Ner, BA,1,2 Mohamad O. Mhajna, MD,1 Asaf Achiron, MD,3 Soad Hajyahia, MD,1 Alon Skaat, MD,1,2 Yakir Berchenko, PhD,4 Bernice Oberman, MSc,4 Ofra Kalter-Leibovici, MD,2,5 Laurence Freedman, PhD,4 Ygal Rotenstreich, MD1,2 Purpose: To assess visual field (VF) defects and retinal function objectively in healthy participants and patients with retinitis pigmentosa (RP) using a chromatic multifocal pupillometer. Design: Cross-sectional study. Participants: The right eyes of 16 healthy participants and 13 RP patients. Methods: Pupil responses to red and blue light (peak, 485 and 625 nm, respectively) presented by 76 light-emitting diodes, 1.8-mm spot size at different locations of a 16.2 VF were recorded. Subjective VFs of RP patients were determined using chromatic dark-adapted Goldmann VFs (CDA-GVFs). Six healthy participants underwent 2 pupillometer examinations to determine testeretest reliability. Main Outcome Measures: Three parameters of pupil contraction were determined automatically: percentage of change of pupil size (PPC), maximum contraction velocity (MCV; in pixels per second), and latency of MCV (LMCV; in seconds). The fraction of functional VF was determined by CDA-GVF. Results: In healthy participants, higher PPC and MCV were measured in response to blue compared with red light. The LMCV in response to blue light was relatively constant throughout the VF. Healthy participants demonstrated higher PPC and MCV and shorter LMCV in central compared with peripheral test points in response to red light. Testeretest correlation coefficients were 0.7 for PPC and 0.5 for MCV. In RP patients, test point in which the PPC and MCV were lower than 4 standard errors from the mean of healthy participants correlated with areas that were indicated as nonseeing by CDA-GVF. The mean absolute deviation in LMCV parameter in response to the red light between different test point was significantly higher in RP patients (range, 0.16e0.47) than in healthy participants (range, 0.02e0.16; P < 0.0001) and indicated its usefulness as a diagnostic tool with high sensitivity and specificity (area under the receiver operating characteristic curve (AUC), 0.97, ManneWhitneyeWilcoxon analysis). Randomly reducing the number of test points to a total of 15 points did not significantly reduce the AUC in RP diagnosis based on this parameter. Conclusions: This study demonstrates the feasibility of using a chromatic multifocal pupillometer for objective diagnosis of RP and assessment of VF defects. Ophthalmology 2016;-:1e14 ª 2016 by the American Academy of Ophthalmology.
Visual field (VF) testing is part of the current clinical standard for evaluating retinal degeneration and optic nerve damage.1,2 Dark-adapted Goldmann perimetry and automated perimetry are used most commonly for detecting and monitoring patients with retinitis pigmentosa (RP).2,3 These methods bear significant limitations because they are subjective by nature and rely heavily on subject cooperation and attention. Hence, testing of young children, the elderly, and individuals with impaired communication skills is doomed to yield unreliable results. These tests also may be stressful for patients because they need to make conscious decisions on identification of nearthreshold stimuli that appear rapidly and disappear.4 Moreover, test results may be affected by the patient’s fatigue, wakefulness, and attentiveness during the long 2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
procedure. Therefore, constant monitoring and instruction of participants by qualified personnel are needed to obtain reliable results.4 Furthermore, testeretest variability, in particular in peripheral locations and in regions of VF deficits, makes it difficult to determine whether the VF is worsening over the course of serial examinations.5e8 Hence, frequent examinations are needed and misdiagnosis of early stages is common.7,9e11 Several attempts have been made to establish objective perimetry based on pupil light response to perimetric light stimuli. Harms12 was the first to report pupilloperimetry in 1949. More studies followed using perimetric white-light stimuli with some success. However, specificity and sensitivity were not sufficiently high to be of clinical use.13e23 More recently, multifocal pupillographic perimetry using http://dx.doi.org/10.1016/j.ophtha.2016.05.038 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 white and chromatic stimuli arranged in a dartboard pattern was developed (nuCoria Field Analyzer; nuCoria Pty. Ltd., Acton, Australia).24e27 This method analyzes both eyes simultaneously, and although it is promising, it cannot differentiate between the rod and cone systems. Retinitis pigmentosa encompasses a group of progressive retinal degeneration diseases that predominantly affect the rod photoreceptor system, resulting in night blindness in the early phase of the disease and loss of peripheral vision that progresses to tunnel vision. In later stages of RP, degeneration of cone photoreceptors causes progressive decline of visual acuity.28 Disease progression is monitored by electroretinography and perimetry.3,29,30 However, poor testeretest repeatability in patients with RP, specifically in areas with VF deficits,5,6,30,31 limits the ability to assess disease progression and particularly to design and interpret clinical trials of potential therapeutic agents. We recently demonstrated a proof of concept for using a chromatic multifocal pupillometer for detection of VF defects in RP patients. The pupil responses of healthy volunteers and RP patients were recorded at 13 different locations of the 30 VF in response to blue- and red-light stimuli (peak, 485 and 640 nm, respectively; light intensity, 40 cd/m2; target size, 64 mm2).32 Retinitis pigmentosa patients demonstrated a significantly reduced percentage of pupil contraction (PPC) compared with healthy participants in testing conditions that emphasized rod contribution (blue light) in nearly all VF locations. By contrast, the PPC in responses to red light (which emphasizes cone contribution) was reduced significantly in RP patients compared with healthy participants, mostly in peripheral locations. In central locations, there was no significant difference between the PPC of RP patients and healthy participants in response to red light. In a second study, we demonstrated that RP patients demonstrated significantly lower PPC in response to blue light in peripheral locations of the central VF than healthy participants.33 Furthermore, in both studies, minimal PPC was recorded in RP patients in areas that were not detected in dark-adapted chromatic Goldmann perimetry.32,33 In these studies, we evaluated only a single parameter of the pupil response, the PPC. These studies suggested that VF defects as well as rod and cone function may be assessed in RP patients using a chromatic multifocal pupillometer. In the current study, we examined the dynamics of the pupil response in the central VF of RP patients and healthy participants. We analyzed additional parameters of the pupil light response, the maximal contraction velocity (MCV), and the latency of MCV (LMCV) to determine the effect of retinal degeneration on these parameters. To the best of our knowledge, this is the first report of evaluation of the LMCV parameter in pupillometry studies. The central VF of RP patients was assessed using a chromatic multifocal pupillometer and was compared with the patients’ chromatic dark-adapted Goldmann VF (CDAGVF)2 results as well as the pupillometry results of healthy participants. We report that RP patients demonstrated significantly lower PPC and MCV in areas that were reported as nonseeing by CDA-GVF and that the mean
2
absolute deviation in the LMCV parameter between different test point locations was significantly higher in RP patients and may present a valuable parameter as a diagnostic tool for RP.
Methods Participants The Sheba Medical Center Institutional Review Board Ethics Committee approved this trial. The study was conducted according to the tenets of the Declaration of Helsinki and was registered at www.clinicaltrials.gov (identifier, NCT02014389). Informed written consent was obtained from all participants. Sixteen healthy volunteers, age matched with patients (see below; 6 men and 10 women; mean age standard deviation, 38.415.6 years; range, 26e77 years) were included in the study. Inclusion criteria were normal eye examination results, best-corrected visual acuity of 20/20, normal color vision, no history of or current ocular disease, no use of any topical or systemic medications that could adversely influence efferent pupil movements, and normal 24-2 Swedish interactive threshold algorithm results, developed for the Humphrey standard perimeter (Humphrey Field Analyser II, Swedish interactive threshold algorithm 24-2; Carl Zeiss Meditec, Inc., Jena, Germany). The study patient group comprised 13 patients with RP (3 women and 10 men; mean age standard deviation, 36.1514.6 years; range, 20e65 years). Inclusion criteria for RP patients were typical abnormal fundus appearance and previously recorded electroretinography results that were abnormal under scotopic or photopic conditions or both (in compliance with the protocol of the International Society for Clinical Electrophysiology of Vision)34 and typical abnormal kinetic chromatic Goldmann test results (loss of VF that is either concentric or that began superiorly and subsequently demonstrated an arcuate scotoma that progressed either from the nasal or the temporal side; incomplete midperipheral ring scotoma that broke through into the periphery; or a residual central VF, with blue and red isopters that were either superimposed or in which the isopter in response to the red stimulus was larger than that in response to the blue stimulus2,3). Exclusion criteria were a concurrent ocular disease and any other condition affecting the pupil response to light. Data recorded for all patients included gender, diagnosis, and electroretinography responses. Patients were tested for best-corrected visual acuity and for color vision by the Farnsworth D15 test. The right eyes of both healthy and RP participants were examined.
Light Stimuli Light stimuli were presented using a Ganzfeld dome apparatus (Accutome, Inc, Malvern, PA; Fig 1) placed 330 mm from the patient’s eye. All tests were performed in a dark room. The untested eye was covered. Participants were asked to fixate on a white-light fixator (0.9 cd/m2; Fig 1B, white arrow) at the center of the dome. Stimuli were presented from 76 targets (lightemitting diodes) with diameter of 1.8 mm2 in a VF of 16.2 . The wavelength and intensity of light stimuli selected for this study were 6255 nm and 1000 cd/m2 for long-wavelength stimuli (red light) and 4855 nm and 200 cd/m2 for short-wavelength stimuli (blue light). The light intensities were chosen after preliminary calibrations that enabled us to identify the minimal stimulus intensity that yielded a substantial pupil response in 5 healthy participants. Background luminance was 0.05 cd/m2. Light intensities were determined by measurement with the LS-100
Chibel et al
Chromatic Multifocal Pupillometry
Figure 1. Photographs of the chromatic multifocal pupillometer. A, Side view of the device. B, Front view of the device. The black arrow points to the infrared camera pin hole, and the white arrow points to the fixator light-emitting diode. C, The user interface for the chromatic multifocal pupillometer. D, Pupillometer-measured parameters.
luminance meter (Konica Minolta, Tokyo, Japan). Stimulus duration was 1 second and the interstimulus interval was 4 seconds.
Pupil Measurement Pupil diameter was recorded in real time by a computerized infrared high-resolution camera (the camera pinhole is marked with a black arrow in Fig 1B) that recorded the pupil diameter at a frequency of 30 Hz. The software (Accutome, Inc.) searched for the pupil in every image. Participants were dark adapted for 3 minutes before testing. Tests in which the subject blinked during the first 2.5 seconds after stimulus onset were excluded automatically, and the targets were retested.
Analysis of Pupil Responses During the recording of the pupil diameter in response to each light stimulus, 5 parameters were calculated within the software using the change in pupil diameter over time: the initial diameter of the pupil (in pixels), the minimum pupil diameter (in pixels), PPC, MCV (in pixels per second), and LMCV (in seconds). The PPC was determined using the following formula, as described previously32: PPC ¼ ðInitial Pupil Diameter Minimum Pupil DiameterÞ=ðInitial Pupil DiameterÞ 100 The MCV was determined by calculating the maximum rate at which the pupil contracted from the light stimulus between the initial pupil diameter measurement and the minimum pupil diameter measurement. The LMCV was determined by calculating the time point for each pupil response at which the maximum rate of pupil contraction (MCV) occurred from each light stimulus.
Chromatic Dark-Adapted Visual Field Retinitis pigmentosa patients were tested for kinetic VF by CDAGVF.2 Briefly, a Goldmann perimeter (940-ST; Haag-Streit AG, Liebefeld, Switzerland) was used to map patients’ conventional and 2-color dark-adapted VFs. Patients were dark adapted for 30 minutes before testing. The setting used for stimuli were II3c for the long-wavelength stimulus and 2 log units lower in luminance (II3 eC) for the short-wavelength stimulus. For quantification of functional CDA-GVF, a schematic representation of the pupillometer target points was overlaid on the CDA-GVF test record, and pupillometer targets that were within seeing areas by the CDAGVF were scored as 1. Pupillometer targets that were in nonseeing areas by the CDA-GVF were scored as 0. The fraction of functioning VF was calculated as the sum of scores divided by 76.
Statistical Analysis Statistical analyses were performed using Excel (Microsoft, Redmond, WA) and R software version 3.0.1 (The Free Software Foundation [FSF]). Results were presented as mean standard error (SE). Student t test was used to evaluate demographic differences between patients and controls. Testeretest reliability of the pupil response measurements was calculated using Pearson correlation, and the correlation between pupillometer recordings and CDA-GVF testing were calculated with Spearman r test. Variability in LMCV recordings was measured by the mean absolute deviation, and this measure was compared between the study group and the control group using a 2-sided ManneWhitneyeWilcoxon test. The effects of using fewer test points and the robustness of the LMCV for discrimination was examined via simulation by randomly selecting n test points (n ¼ 5, 10, 15, . 75) and calculating the area under the receiver operating characteristic curve (AUC) obtained using the LMCV
3
Ophthalmology Volume -, Number -, Month 2016
Figure 2. Visual field results showing the mean (A, B) percentage of change of pupil size (PPC), (C, D) maximum contraction velocity (MCV), and (E, F) latency of maximum contraction velocity (LMCV) parameters of healthy participants in response to (A, C, E) red and (B, D, F) blue light at each test target location, indicted by number. Color coding for each parameter is shown on the right. deg ¼ degree.
based on these n points. This was done for 200 repeats, and the mean AUC for each n thus was obtained.
Results Characterization of Pupil Responses of Healthy Participants First we characterized the pupil response parameters in each test point for red- and blue-light stimuli in control participants. Figure 2 demonstrates color-coded maps of mean PPC (Fig 2A, B), MCV (Fig 2C, D), and LMCV (Fig 2E, F) recorded from control participants in each test point location in response to red-light stimuli (Fig 2A, C, E) and in response to blue-light stimuli (Fig 2B, D, F). The mean PPC recorded in response to blue-light stimuli ranged from 13% to 28% at different test point locations
4
(mean SE, 19.40.22%) and was significantly higher compared with the red-light stimuli (range, 7%e22%; mean SE, 120.2%; P < 0.0001, t test), even though the red-light stimuli were presented at a 5-fold higher intensity than the blue-light stimuli. Similarly, the mean MCV was significantly higher in response to blue-light stimuli (range, 28e43 pixels/second; mean SE, 370.3 pixels/second) compared with the red-light stimuli (range, 18e36 pixels/second; mean SE, 23.173 pixels/second; P ¼ 2107, t test). There was no significant difference in the mean LMCV between the red and blue light (range, 0.6e0.8 seconds; P ¼ 0.11, t test). Higher PPC was recorded in central locations of the VF compared with peripheral locations in response to both red and blue light (Fig 2A, B). A similar pattern of faster mean MCV in central locations compared with peripheral locations was demonstrated clearly in response to red-light stimuli (Fig 2C). The centraleperipheral gradient pattern was less evident in
Chibel et al
Chromatic Multifocal Pupillometry
Figure 3. Scatterplots showing the test versus retest of (A, B) percentage of change of pupil size (PPC), (C, D) maximum contraction velocity (MCV), and (E, F) latency of maximum contraction velocity (LMCV) parameters in serial testing of 6 healthy participants in response to (A, C, E) blue and (B, D, F) red light. sec ¼ second.
response to blue light (Fig 2D). The LMCV parameter was relatively constant throughout the VF field in response to both red and blue light, with an interquartile range of 0.6 to 0.7 seconds. The longest LMCV values were measured at peripheral test points in response to both wavelengths (Fig 2E, F). Test reliability was assessed by retesting 6 healthy controls. Good linear correlation between the test and the retest was demonstrated for the parameter PPC for both colors (blue: R2 ¼ 0.762; P < 0.0001; red: R2 ¼ 0.721; P < 0.0001; Fig 3A, B). The parameter MCV demonstrated lower but still reasonable correlation between test and retest (blue: R2 ¼ 0.513; P < 0.0001; red: R2 ¼ 0.522; P < 0.0001; Fig 3C, D). The lowest correlation between test and retest was recorded for the LMCV parameter (blue: R2 ¼ 0.419; P < 0.0001; red: R2 ¼ 0.208; P < 0.0001; Fig 3E, F).
Retinitis Pigmentosa Patients Demonstrated Diminished Pupil Responses in Correlation with Disease Severity Thirteen RP patients were divided into 2 groups based on their CDA-GVF testing results. Group A consisted of 5 patients with
some functional CDA-GVF in response to red and blue light, whereas group B included 8 patients with a severe disease with no detection of either blue or red light by CDA-GVF (Table 1). Figure 4 demonstrates grey-scale maps of mean PPC (Fig 4AeC), MCV (Fig 4DeF), and LMCV (Fig 4GeI) recorded in each test point location in response to blue light in healthy participants (Fig 4A, D, G) and in patients with intermediate retinal degeneration (group A; Fig 4B, E, H) and severe retinal degeneration (group B; Fig 4C, F, I). Color coding was set to resemble the Humphrey perimeter’s output, with a white color for normal values and darker colors for values that were less than normal. Normal values were set as the mean of healthy participants in each test point location (Figs 2 and 4). Deviation from normal values was determined based on the SEs calculated for each parameter in each target point in the healthy participants. Thus, for PPC and MCV, the darkest color was used for test points in which the mean of patients was less than 5 SEs from the mean of healthy participants in those points. For LMCV, the darkest color was used for test points in which the mean of patients was more than 5 SEs from the mean of healthy participants in those points. In group B, mean PPC and MCV in response to the blue light were less than 5 SEs from the mean of healthy participants in nearly all test point locations (Fig 4C, F).
5
ND
F ¼ female; M ¼ male; ND ¼ not detected; RP ¼ retinitis pigmentosa; US ¼ Usher syndrome. *Group A ¼ some light detection in response to both red and blue by chromatic dark-adapted Goldmann visual field; group B ¼ no light detection in response to at least 1 color by chromatic dark-adapted Goldmann visual field.
ND 33 23 30 26 46 ND ND 38 26 ND ND ND ND 26 33 5 5 13 ND ND 22 10 ND ND ND ND 16 6 ND 31 16 ND ND 26 17 ND ND ND ND 16 21 ND 9 7 ND ND 9 10 ND ND ND ND 39 42 36 51 44 ND ND 57 44 ND ND ND ND 24.5 17 ND 30 19 ND ND 14 17 ND ND ND ND 95 15 ND 16 15 ND ND 6 15 ND
ND 100 ND ND 78 ND ND ND ND ND ND ND ND ND 74 ND ND 49 ND ND ND ND ND ND ND ND B A A A A A B B B B B B B 20/40 20/3600 20/25 20/32 20/3600 20/20 20/50 20/63 20/64 20/640 20/20 20/30 20/50 RP RP RP RP US RP RP RP RP RP RP RP RP 34 44 34 32 20 31 27 55 21 22 28 65 58 M M F F M F M M M M M M M 1 2 3 4 5 6 7 8 24 25 26 29 30
Group* Gender
Age (yrs)
Diagnosis
Best-Corrected Visual Acuity Patient No.
6
ND 149 11 5 34 7 ND ND 26 17 ND ND ND
Implicit Time (ms) Maximum (mm) Implicit Time (ms)
Single-Flash A-Wave
Maximum (mm) Implicit Time (ms)
Dark-Adapted B-Wave
Maximum (mm) Implicit Time (ms)
Dark-Adapted A-Wave
Maximum (mm) Implicit Time (ms)
Isolated Rod Response
Electroretinography Results
Table 1. Summary of Patient Characteristics
Maximum (mm)
Single-Flash B-Wave
Ophthalmology Volume -, Number -, Month 2016 Similarly, the mean LMCV was more than 5 SEs from the mean of healthy participants in 68 of 76 test points (89% of the VF; Fig 4I). By contrast, the mean PPC recorded in patients in group A was equal to or was lower by less than 2 SEs from the mean of healthy participants in most of the VF test points (50 test points, 66% of the VF; Fig 4B). In 26 test points (34% of the VF) that were located mostly in the periphery of the VF, the mean PPC of group B patients was less than 2 SEs from the mean of healthy participants. The MCV and LMCV parameters also demonstrated an intermediate defect in pupil response. Thus, in 33 test point locations (43% of the VF), the mean MCV was less than 3 SEs from the mean of healthy participants (Fig 4E), and in 25 test point locations (33% of the VF), the LMCV was less than 3 SEs from the mean of healthy participants (Fig 4E, H). Figure 5 demonstrates the pupil responses to red light in healthy participants and RP patients. Retinitis pigmentosa patients from both groups demonstrated lower PPC and MCV and longer LMCV compared with healthy participants, but to a smaller extent than the response to the blue light. Thus, in group B, the mean PPC and MCV were less than 5 SEs from the mean of healthy participants in 35 (46% of the VF) and 57 (75% of the VF) test points, respectively (Fig 5C, F). Similarly, the mean LMCV was more than 5 SEs from the mean of healthy participants in 50 test points (66% of VF) compared with 68 points (89% of the VF) in response to blue light (Fig 5I). Group A demonstrated a milder decline in pupil responses to red light compared with group B, with only 4 (5% of the VF) and 9 (12% of the VF) test points in which the PPC and MCV were less than 5 SEs from the mean of healthy participants, respectively (Fig 5B, E). The mean LMCV was more than 5 SEs from the mean of healthy participants in 35 test points (46% of the VF; Fig 5H).
Variability in Latency of Maximum Contraction Velocity as a Diagnostic Tool for Retinitis Pigmentosa As shown in Figures 2 and 4, LMCV was relatively constant in healthy participants in response to blue and red light in most test point locations, ranging from 0.6 to 0.8 seconds. By contrast, this parameter was highly variable between different test point locations in RP patients, ranging from 0.6 to 1.7 seconds and from 0.6 to 1.4 seconds in response to the blue and red light, respectively (Figs 4 and 5). To evaluate the extent of the variability in LMCV between different test point locations of the VF, the mean response for each subject was determined (i.e., the mean LMCV among the subject’s 76 test points). Then, the mean absolute deviation was calculated as the mean of the absolute difference between the mean and the measurements in each of the test points. Figure 6A demonstrates for each participant a boxplot depicting the distribution of the LMCV parameter for all test points in response to the red light. The mean absolute deviation in LMCV in response to the red light between different test points of each participant was significantly higher in patients with RP than in healthy participants (P < 106, ManneWhitneyeWilcoxon test). In addition, the ManneWhitneyeWilcoxon statistic test indicated that a classification method based on measurement of LMCV in response to red light would have an AUC of 0.97. Similarly, the mean absolute deviation of LMCV in response to the blue light between different test points of each participant was significantly higher in patients with RP than in healthy participants (P ¼ 104, WilcoxoneManneWhitney test; AUC, 0.93; data not shown). There was no significant difference in the absolute mean
Chibel et al
Chromatic Multifocal Pupillometry
Figure 4. Visual field (VF) results showing the mean (AeC) percentage of change of pupil size (PPC), (DeF) maximum contraction velocity (MCV), and (GeI) latency of maximum contraction velocity (LMCV) parameters recorded in response to blue light in (A, D, G) healthy participants and in patients from (B, E, H) group A and (C, F, I) group B in each of the 76 test targets of the 16.2 VF. Color coding is based on number of standard errors (SEs) from the mean of healthy participants and is shown on the right. White ¼ mean value equals or is less than 1 SE from the mean of healthy participants in each location; dark gray ¼ mean values differs in more than 5 SEs from the mean of healthy participants in each location. deg ¼ degree.
deviation of the PPC and MCV parameters between RP patients and healthy controls (data not shown). The mean absolute deviation in LMCV in response to the red light correlated negatively with the fraction of functional subjective VF as determined by CDA-GVF (Spearman r, 0.45; P ¼ 0.13; Fig 6B). The highest mean absolute deviation in LMCV (>0.3) was found in patients with no light detection by CDA-GVF. By contrast, lower mean absolute deviation in LMCV (<0.3) was demonstrated in patients with some functional CDA-GVF.
Clustering Analysis Next we examined whether fewer test point locations could be used for RP diagnosis in an attempt to optimize the analysis of pupillometer-based perimetry and reduce testing time. Figure 7 shows the mean and standard deviation of the AUC obtained after a random selection of test locations. Randomly reducing the number of test points to a total of 15 points did not significantly reduce the AUC in RP diagnosis based on the absolute mean deviation of LMCV. The AUC of 0.90 (highlighted by a dashed
red line) suggests that the probability of LMCV to discriminate successfully a (randomly selected) RP patient from a (randomly selected) healthy subject is 0.9.
Individual Retinitis Pigmentosa Cases To illustrate the pattern of recorded pupil response values compared with the results of subjective CDA-GVF testing, the individual reports for 3 RP patients are presented in Figures 8, 9, and 10. Patient 30: No Light Detection by Chromatic Dark-Adapted Goldmann Visual Field. Patient 30 showed no light detection by CDA-GVF (Fig 8A, E). The PPC and MCV parameters were less than 5 SEs from the mean of healthy participants in 75 of the 76 test points (99% of the VF) in response to blue light (Fig 8B, C). The LMCV parameter was more than 5 SEs from the mean of healthy participants in 48 of the 76 test points (63% of the VF; Fig 8D). The pupil responses to red light also were diminished significantly throughout the VF, with 69 and 74 of the 76 test points (91%
7
Ophthalmology Volume -, Number -, Month 2016
Figure 5. Visual field (VF) results showing the mean (AeC) percentage of change of pupil size (PPC), (DeF) maximum contraction velocity (MCV), and (GeI) latency of maximum contraction velocity (LMCV) parameters in response to red light in (A, D, G) healthy participants and in patients from (B, E, H) group A and (C, F, I) group B in each of the 76 test targets of the 16.2 VF. Color coding is based on number of standard errors (SEs) from the mean of healthy participants and is shown on the right, as described in Figure 4. deg ¼ degree.
and 77% of the VF), respectively, presenting PPC and MCV less than 4 SEs from the mean of healthy participants (Fig 8F, G). The LMCV parameter was more than 5 SEs from the mean of healthy participants in 38 test points (50% of the VF; Fig 8H). The mean absolute deviation in LMCV in response to the red light for this patient was the largest recorded in this study (0.47 second; Fig 6A). Patient 4: Tunnel Vision by Chromatic Dark-Adapted Goldmann Visual Field. Patient 4 demonstrated tunnel vision on CDA-GVF (Fig 9A, E). The map of the PPC parameter in response to blue light correlated well with the CDA-GVF map. Thus, in peripheral test point locations (nonseeing by CDA-GVF), the PPC values were less than 5 SEs from the mean of healthy participants, whereas in central locations, PPC values were only 1 to 3 SEs less than the mean of healthy participants (Fig 9B). The map of MCV demonstrated substantial reduction of MCV throughout the VF, with MCV less than 4 SEs from the mean of healthy participants in all 76 test points (100% of the VF; Fig 9C). The LMCV parameter was 1 to 2 SEs from the mean of healthy participants in 20 test points (26% of the VF), mostly in central locations (Fig 9D).
8
A similar tunnel vision pattern of pupil responses was obtained in response to the red light. The PPC and MCV parameters were equal to or only 1 SE from the mean of healthy participants in 14 and 8 central test point locations, correlating to 18% and 11% of the VF, respectively. By contrast, the PPC and MCV recorded in nearly all peripheral test points were more than 5 SEs less than the mean of healthy participants (Fig 9F, G). The LMCV parameter was close to normal (1e2 SEs from the mean of healthy participants) in 38 test points (50% of the VF). Most of these test points were located at the center of the VF (Fig 9H). The mean absolute deviation in LMCV in response to the red light for this patient was intermediate (0.26 second; Fig 6A). Patient 3: Larger Fraction of Functional Chromatic DarkAdapted Goldmann Visual Field. Patient 3 had a significant functional portion of the 16.2 VF as determined by the subjective CDA-GVF (Fig 10A, E). The pupil responses of this patient to both the red and blue light were equal to or 1 to 2 SEs from the mean of healthy participants in most test points located in functional areas determined by the CDA-GVF. Most areas that demonstrated substantially reduced PPC and MCV and longer LMCV compared with control values (more than 2 SEs from the mean of healthy
Chibel et al
Chromatic Multifocal Pupillometry
Figure 6. The variability in parameter latency of maximum contraction velocity (LMCV) in response to red-light stimuli between different visual field (VF) test point locations in retinitis pigmentosa (RP) patients correlates with their chromatic dark-adapted Goldmann VF (CDA-GVF) results. A, Box-and-whisker plot showing that RP patients demonstrate significantly higher variability in LMCV in response to red-light stimuli between different VF locations compared with control subjects. The absolute difference between the mean LMCV (of the 76 points) and the recordings in 76 points was determined for each subject (y-axis). x-axis ¼ subject number. Boxes represent the first and third quartiles and the band inside the box represents the second quartile (the median). The whiskers extend to the most extreme data point, which is no more than 1.5 times the interquartile range from the box. B, Graph showing negative linear correlation between the mean absolute deviation in LMCV in response to red light and CDA-GVF. y-axis ¼ the mean absolute deviation of LMCV (of the 76 points) in response to red light. x-axis ¼ the fraction of functioning VF in response to red light that was calculated for RP patients based on their subjective CDA-GVF results. Solid line ¼ linear regression.
participants) were located mostly in nonseeing areas by CDA-GVF (Fig 10). The mean absolute deviation in LMCV in response to the red light for this patient was small (0.16 second; Fig 6A).
Discussion This study demonstrated the feasibility of using a chromatic multifocal pupillometer for objective assessment of VF
Figure 7. The computational random selection analysis of the mean absolute deviation in latency of maximum contraction velocity parameter: the mean and standard deviation of the area under the receiver operating characteristic curve (AUC) obtained after a random selection of test locations (y-axis) versus the number of test points used (x-axis).
defects in patients with RP. Test point locations in which PPC and MCV were less than 4 SEs from the mean of healthy participants correlated with areas that were abnormal (nonseeing) by CDA-GVF. The RP patients with severe VF loss presented more testing points that differed substantially from the mean of healthy participants compared with patients with a moderate loss of VF, particularly in response to the blue-light stimuli. Patients with some functional VF demonstrated reduced pupil light response (PLR) particularly in peripheral test points and in response to the blue-light stimuli. The pathologic features of RP are characterized by a loss of rod function that exceeds the reduction of cone function, and the VF loss typically begins with peripheral VF constriction.3,28 Our finding that pupil response to blue light in RP patients was more affected than the pupil response to red-light stimuli strongly suggests that the pupil response to blue light measured by our chromatic pupillometer in response to focal light stimuli is mediated mainly by rods, whereas the pupil response to red light is mediated mainly by cones. Hence, a chromatic multifocal pupillometer may enable objective noninvasive assessment of the function of rods and cones at distinct locations of the VF for the first time. These results are supported by our previous study that demonstrated reduced pupil responses in peripheral VF locations in response to blue light32 and by the studies from Kawasaki et al,35 Kardon et al,36,37 and Lorenz et al38 that demonstrated that transient pupil response to full-field low-intensity blue-light stimulus reflects rod activity and that transient pupil response to red light is driven predominantly by cones. Curcio et al39 reported that the average horizontal diameter of the rod-free zone in the human fovea is 0.350 mm (1.25 ).
9
Ophthalmology Volume -, Number -, Month 2016
Figure 8. A, E, Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in patient 30 demonstrated no light detection. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of maximum contraction velocity (LMCV) parameters of this patient in response to (BeD) blue-light and (FeH) red-light stimuli recorded in each of the 76 points of the 16.2 visual field are presented. Color coding is the same as described in Figure 4. deg ¼ degree.
Because the 4 most central pupillometer test point locations are located at 1.85 , the pupil responses to blue light recorded from these 4 central test points may be mediated by foveal rods. Since the pupil responses recorded in this study were transient and because dim light was used in the current test protocol, it is less likely that intrinsically photosensitive retinal ganglion cells are involved in pupil response to blue-light stimuli presented at the 4 central test points. However, because cones are highly concentrated in the fovea, it is possible that some of the pupil response to blue light in the 4 central test points is mediated by S cones. We plan to examine this possibility in a future study by testing coneerod dystrophy patients. The intensity of blue-light stimulus used in our study was 5-fold lower than that of the red-light stimulus. Nevertheless, pupil responses to blue-light stimulus in healthy participants were stronger than the responses to red-light stimuli in the same test point locations (Fig 2). These findings correlate with those of our previous study using the first chromatic pupillometer prototype32 and with the findings of Kawasaki et al,35 Kardon et al,36,37 and Lorenz et al,38 and may be explained by the lower number of cones compared with rods in the human retina39 and by the smaller
10
receptive fields of cones and their lower sensitivity for light compared with rods. The LMCV parameter recorded in response to red-light stimuli seems to be a useful tool for noninvasive and objective diagnosis of RP with an AUC of 0.97. Our computer-based random clustering analysis suggested that shortening test duration may be possible with computational clustering, without reducing the sensitivity and accuracy of RP diagnosis. Thus, testing only 15 test points in response to red-light stimuli, which is predicted to take 1 minute, would enable diagnosis of RP with an AUC of 0.9. Importantly, our study group included RP patients at different stages of the diseases, some with substantial functional 16.2 VF (such as patient 3) and some patients who had no light detection (such as patient 30), further emphasizing the high specificity and sensitivity of the LMCV score. In coming clinical trials using a chromatic multifocal pupillometer, we will test a larger VF (30 ) using a larger cohort of patients and healthy participants. The future study also will include more thorough computational optimization for clustering based on the promising results of random reduction of target locations. In addition, future trials are planned in which the pupillometry testing of patients at early stages and healthy participants will be
Chibel et al
Chromatic Multifocal Pupillometry
Figure 9. A, E, Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in patient 3 demonstrating the subjective visual field (VF) for (A) blue and (E) red light. The dashed green line marks the borders of the 16.2 VF. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of MCV (LMCV) parameters of the patient in response to (BeD) blue-light and (FeH) red-light stimuli recorded in each of the 76 points of the 16.2 VF are presented. Color coding is the same as described in Figure 4. deg ¼ degree.
performed in a masked fashion to examine the applicability of pupillometry testing as a diagnostic tool. Previous studies suggested direct correlation between pupil response amplitude, maximum velocity, and latency in response to full-field white-light stimulus. The authors noted that most of the latency in healthy participants was the result of delay in iris smooth muscle contraction and that a relatively small part is the result of conduction along the pupil reflex pathway.40 Because apoptosis in the iris muscle increases with age, we included agematched groups with a wide range of ages (26e77 years). To the best of our knowledge, our study enables mapping different pupil light response parameters in different locations of the VF in response to red and blue light for the first time. Our findings suggest that in most cases, maximum pupil contraction velocity correlated directly with change in pupil diameter. However, some exceptions were noted (for example, in patient 4). A future clinical trial with a larger study group may enable determination of the correlation between the different pupil response parameters in this new setting of response to focal chromatic light stimuli. One of the major limitations of subjective perimetry is testeretest variability. A good correlation was demonstrated in PPC and MCV in testeretest analysis. The large bin size
of the LMCV parameter (0.033 second) and outliers may be the cause for the low testeretest correlation of LMCV. A future clinical trial will examine the test reliability in repeated testing of a larger number of healthy participants and patients. The chromatic multifocal pupillometer device presented here was built to enable mapping of the central VF. We chose to examine the device in RP patients because of their pathologic features that facilitate the differentiation between cone and rod function. Future clinical trials will include patients with other blinding diseases, such as patients with macular degeneration and glaucoma, using a device that enables 30 VF testing. The major limitation of this study is the size of the study cohorts. This is a preliminary study that is being followed by a larger clinical trial with more types of diseases. Several studies demonstrated a correlation between structural abnormalities on optical coherence tomography (OCT), including decreased outer nuclear layer and outer segment thickness and the loss of local VF sensitivity of RP patients.41e44 Wen et al45 demonstrated strong positive correlation between cone function preservation measured by multifocal electroretinography, VF sensitivity, and the remaining thickness of the photoreceptor layer by spectral-domain OCT in RP patients. Recent adaptive
11
Ophthalmology Volume -, Number -, Month 2016
Figure 10. Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in retinitis pigmentosa patient 4 demonstrating the subjective visual field (VF) for (A) blue and (E) red light. The dashed green line marks the borders of the 16.2 VF. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of MCV (LMCV) parameters of this patient in response to (BeD) blue-light and (FeH) redlight stimuli recorded in each of the 76 points of the 16.2 VF are presented. Color coding is the same as in Figure 4. deg ¼ degree.
optics scanning laser ophthalmoscopy studies demonstrated early detection of reduced cone density and irregular cone mosaic spatial arrangement in the macula of RP patients with best-corrected visual acuity of 20/20.46e48 Future trials will include follow-up pupillometry and spectral-domain OCT imaging that will allow analysis of longitudinal changes in pupillary response with respect to disease progression and examination of the correlation between functional pupillometry findings and spectraldomain OCT structural findings. In addition, it would be interesting to examine the correlation between the chromatic multifocal pupillometry function testing and adaptive optics scanning laser ophthalmoscopy imaging. In the current study, the pupillometer test results were compared with CDA-GVF that enables evaluation of cone and rod responses. In future studies, the pupillometer results will be compared with the more commonly used Humphrey perimetry (Carl Zeiss Meditec, Inc., Jena, Germany). Taken together, our results suggest direct correlation between seeing and pupil response function of the retina. Different parameters of pupil response can be used for VF determination (PPC and MCV) and disease diagnosis (stimulus color and LMCV). The pupillometer test is predicted to be less stressful for tested subjects because patients are unaware of the test results and the test may facilitate
12
objective perimetry and assessment of function of retinal photoreceptors with minimal patient cooperation and minimal technician training.
References 1. Sharma P, Sample PA, Zangwill LM, et al. Diagnostic tools for glaucoma detection and management. Surv Ophthalmol 2008;53(suppl 1):S17–32. 2. Rotenstreich Y, Fishman GA, Lindeman M, et al. The application of chromatic dark-adapted kinetic perimetry to retinal diseases. Ophthalmology 2004;111:1222–7. 3. Grover S, Fishman GA, Brown J Jr. Patterns of visual field progression in patients with retinitis pigmentosa. Ophthalmology 1998;105:1069–75. 4. Trope GE, Britton R. A comparison of Goldmann and Humphrey automated perimetry in patients with glaucoma. Br J Ophthalmol 1987;71:489–93. 5. Kim LS, McAnany JJ, Alexander KR, et al. Intersession repeatability of Humphrey perimetry measurements in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2007;48: 4720–4. 6. Seiple W, Clemens CJ, Greenstein VC, et al. Testeretest reliability of the multifocal electroretinogram and Humphrey
Chibel et al
7. 8.
9. 10. 11. 12. 13.
14. 15. 16. 17. 18. 19. 20. 21. 22.
23. 24. 25. 26. 27.
Chromatic Multifocal Pupillometry
visual fields in patients with retinitis pigmentosa. Documenta Ophthalmologica 2004;109:255–72. Heijl A, Lindgren A, Lindgren G. Test-retest variability in glaucomatous visual fields. Am J Ophthalmol 1989;108:130–5. Artes PH, Iwase A, Ohno Y, et al. Properties of perimetric threshold estimates from full threshold, SITA standard, and SITA fast strategies. Invest Ophthalmol Visual Sci 2002;43: 2654–9. Turalba AV, Grosskreutz C. A review of current technology used in evaluating visual function in glaucoma. Semin Ophthalmol 2010;25:309–16. Chauhan BC, Garway-Heath DF, Goni FJ, et al. Practical recommendations for measuring rates of visual field change in glaucoma. Br J Ophthalmol 2008;92:569–73. Advanced Glaucoma Intervention Study Group. Advanced Glaucoma Intervention Study 2: visual field test scoring and reliability. Ophthalmology 1994;101:1445–55. Harms H. Grundlagen, methodik und bedeutung der pupillenperimetrie für die physiologie und pathologie des sehorgans. Graefes Arch Clin Exp Ophthalmol 1949;149:1–68. Reuther R, Krastel H, Alexandridis E. Disturbances of the pupil reflex associated with homonymous hemianopic paracentral scotoma [in German]. Arch Psychiatr Nervenkr 1981;229:259–66. Asakawa K, Shoji N, Ishikawa H, et al. New approach for the glaucoma detection with pupil perimetry. Clin Ophthalmol 2010;4:617–23. Hong S, Narkiewicz J, Kardon RH. Comparison of pupil perimetry and visual perimetry in normal eyes: decibel sensitivity and variability. Invest Ophthalmol Vis Sci 2001;42:957–65. Aoyama T. Pupillographic perimetry. the application to clinical cases [in Japanese]. Nihon Ganka Gakkai Zasshi 1977;81: 1527–38. Chang DS, Xu L, Boland MV, et al. Accuracy of pupil assessment for the detection of glaucoma: a systematic review and meta-analysis. Ophthalmology 2013;120:2217–25. Alexandridis E, Krastel H. Pupillographic perimetry using the “octopus”. Fortschr Ophthalmol 1989;86:692–4. Yoshitomi T, Matsui T, Tanakadate A, et al. Comparison of threshold visual perimetry and objective pupil perimetry in clinical patients. J Neuroophthalmol 1999;19:89–99. Schmid R, Luedtke H, Wilhelm BJ, et al. Pupil campimetry in patients with visual field loss. Eur J Neurol 2005;12:602–8. Skorkovska K, Wilhelm H, Ludtke H, et al. How sensitive is pupil campimetry in hemifield loss? Graefes Arch Clin Exp Ophthalmol 2009;247:947–53. Skorkovska K, Ludtke H, Wilhelm H, et al. Pupil campimetry in patients with retinitis pigmentosa and functional visual field loss. Graefes Arch Clin Exp Ophthalmol 2009;247: 847–53. Jensen W. A description of a method for objective perimetry [in German]. Graefes Arch Clin Exp Ophthalmol 1976;201: 183–91. Sabeti F, Maddess T, Essex RW, et al. Multifocal pupillographic assessment of age-related macular degeneration. Optom Vis Sci 2011;88:1477–85. Maddess T, Ho YL, Wong SS, et al. Multifocal pupillographic perimetry with white and colored stimuli. J Glaucoma 2011;20:336–43. Carle CF, James AC, Kolic M, et al. Blue multifocal pupillographic objective perimetry in GlaucomaBlue mfPOP in glaucoma. Invest Ophthalmol Vis Sci 2015;56:6394–403. Sabeti F, Nolan CJ, James AC, et al. Multifocal pupillography identifies changes in visual sensitivity according to severity of diabetic retinopathy in type 2 DiabetesmfPOP to identify
28. 29. 30. 31. 32.
33.
34. 35.
36.
37. 38.
39. 40. 41. 42. 43.
44.
45.
46.
patients with and without DR. Invest Ophthalmol Vis Sci 2015;56:4504–13. Hartong DT, Berson EL, Dryja TP. Retinitis pigmentosa. Lancet 2006;368:1795–809. Birch DG, Anderson JL, Fish GE. Yearly rates of rod and cone functional loss in retinitis pigmentosa and cone-rod dystrophy. Ophthalmology 1999;106:258–68. Holopigian K, Greenstein V, Seiple W, et al. Rates of change differ among measures of visual function in patients with retinitis pigmentosa. Ophthalmology 1996;103:398–405. Ross DF, Fishman GA, Gilbert LD, et al. Variability of visual field measurements in normal subjects and patients with retinitis pigmentosa. Arch Ophthalmol 1984;102:1004–10. Skaat A, Sher I, Kolker A, et al. Pupillometer-based objective chromatic perimetry in normal eyes and patients with retinal photoreceptor dystrophies. Invest Ophthalmol Vis Sci 2013;54:2761–70. Rotenstreich Y, Chibel R, Yahia SH, et al. The first prototype of chromatic pupillometer for objective perimetry in retinal degeneration patients. SPIE BiOS Int Soc Optics Photonics 2015. 93070Qe93070Q-7. Marmor MF, Fulton AB, Holder GE, et al. ISCEV standard for full-field clinical electroretinography (2008 update). Doc Ophthalmol 2009;118:69–77. Kawasaki A, Crippa SV, Kardon R, et al. Characterization of pupil responses to blue and red light stimuli in autosomal dominant retinitis pigmentosa due to NR2E3 mutation. Invest Ophthalmol Vis Sci 2012;53:5562–9. Kardon R, Anderson SC, Damarjian TG, et al. Chromatic pupil responses: preferential activation of the melanopsinmediated versus outer photoreceptor-mediated pupil light reflex. Ophthalmology 2009;116:1564–73. Kardon R, Anderson SC, Damarjian TG, et al. Chromatic pupillometry in patients with retinitis pigmentosa. Ophthalmology 2011;118:376–81. Lorenz B, Strohmayr E, Zahn S, et al. Chromatic pupillometry dissects function of the three different light-sensitive retinal cell populations in RPE65 deficiency. Invest Ophthalmol Vis Sci 2012;53:5641–52. Curcio CA, Sloan KR, Kalina RE, et al. Human photoreceptor topography. J Comp Neurol 1990;292:497–523. Ellis CJ. The pupillary light reflex in normal subjects. Br J Ophthalmol 1981;65:754–9. Rangaswamy NV, Patel HM, Locke KG, et al. A comparison of visual field sensitivity to photoreceptor thickness in retinitis pigmentosa. Invest Ophthalmol Vis Sci 2010;51:4213–9. Hood DC, Lazow MA, Locke KG, et al. The transition zone between healthy and diseased retina in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2011;52:101–8. Jacobson SG, Cideciyan AV, Aleman TS, et al. Usher syndromes due to MY07A, PCDH15, USH2A or GPR98 mutations share retinal disease mechanism. Hum Mol Genet 2008;17:2405–15. Jacobson SG, Aleman TS, Sumaroka A, et al. Disease boundaries in the retina of patients with usher syndrome caused by MY07A gene mutations. Invest Ophthalmol Vis Sci 2009;50:1886–94. Wen Y, Klein M, Hood DC, et al. Relationships among multifocal electroretinogram amplitude, visual field sensitivity, and SD-OCT receptor layer thicknesses in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2012;53:833–40. Makiyama Y, Ooto S, Hangai M, et al. Macular cone abnormalities in retinitis pigmentosa with preserved central vision using adaptive optics scanning laser ophthalmoscopy. PLoS One 2013;8:e79447.
13
Ophthalmology Volume -, Number -, Month 2016 47. Park SP, Lee W, Bae EJ, et al. Early structural anomalies observed by high-resolution imaging in two related cases of autosomal-dominant retinitis pigmentosa. Ophthal Surg Lasers Imaging 2014;45:469–73.
48. Park SP, Hong IH, Tsang SH, et al. Cellular imaging demonstrates genetic mosaicism in heterozygous carriers of an X-linked ciliopathy gene. Eur J Hum Genet 2013;21: 1240–8.
Footnotes and Financial Disclosures Originally received: January 11, 2016. Final revision: May 21, 2016. Accepted: May 23, 2016. Available online: ---.
Author Contributions: Conception and design: Rotenstreich, Sher Manuscript no. 2016-92.
1
The Goldschleger Eye Institute, Sheba Medical Center, Tel-Hashomer, Israel.
2
Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
3
Department of Ophthalmology, Wolfson Medical Center, Holon, Israel. 4 Biostatistics Unit, Gertner Institute for Epidemiology and Health Policy Research, Ramat Gan, Israel. 5
Unit of Cardiovascular Epidemiology, Gertner Institute for Epidemiology and Health Policy Research, Ramat Gan, Israel.
Financial Disclosure(s): The author(s) have made the following disclosure(s): Y.R.: Financial support (to institution) e Accutome Inc., Malvern, PA; Patent e Sheba Medical Center, Tel-Hashomer, Israel Supported by Accutome, Inc., Malvern, PA; Accutome, Inc., participated in review of the manuscript but had no role in the design or conduct of this research.
14
Analysis and interpretation: Chibel, Sher, Achiron, Skaat, Berchenko, Oberman, Kalter-Leibovici, Freedman, Rotenstreich Data collection: Chibel, Sher, Ben Ner, Mhajna, Hajyahia Obtained funding: Rotenstreich Overall responsibility: Rotenstreich Abbreviations and Acronyms: AUC ¼ area under the receiver operating characteristic curve; CDAGVF ¼ chromatic dark-adapted Goldmann visual field; LMCV ¼ latency of maximum contraction velocity; MCV ¼ maximum contraction velocity; OCT ¼ optical coherence tomography; PPC ¼ percentage of change of pupil size; RP ¼ retinitis pigmentosa; SE ¼ standard error; VF ¼ visual field. Correspondence: Ygal Rotenstreich, MD, Goldschleger Eye Institute, Sheba Medical Center, Tel Hashomer 52621, Israel. E-mail: Ygal.Rotenstreich@sheba. health.gov.il.
Visual field (VF) testing is part of the current clinical standard for evaluating retinal degeneration and optic nerve damage.1,2 Dark-adapted Goldmann perimetry and automated perimetry are used most commonly for detecting and monitoring patients with retinitis pigmentosa (RP).2,3 These methods bear significant limitations because they are subjective by nature and rely heavily on subject cooperation and attention. Hence, testing of young children, the elderly, and individuals with impaired communication skills is doomed to yield unreliable results. These tests also may be stressful for patients because they need to make conscious decisions on identification of nearthreshold stimuli that appear rapidly and disappear.4 Moreover, test results may be affected by the patient’s fatigue, wakefulness, and attentiveness during the long 2016 by the American Academy of Ophthalmology Published by Elsevier Inc.
procedure. Therefore, constant monitoring and instruction of participants by qualified personnel are needed to obtain reliable results.4 Furthermore, testeretest variability, in particular in peripheral locations and in regions of VF deficits, makes it difficult to determine whether the VF is worsening over the course of serial examinations.5e8 Hence, frequent examinations are needed and misdiagnosis of early stages is common.7,9e11 Several attempts have been made to establish objective perimetry based on pupil light response to perimetric light stimuli. Harms12 was the first to report pupilloperimetry in 1949. More studies followed using perimetric white-light stimuli with some success. However, specificity and sensitivity were not sufficiently high to be of clinical use.13e23 More recently, multifocal pupillographic perimetry using http://dx.doi.org/10.1016/j.ophtha.2016.05.038 ISSN 0161-6420/16
1
Ophthalmology Volume -, Number -, Month 2016 white and chromatic stimuli arranged in a dartboard pattern was developed (nuCoria Field Analyzer; nuCoria Pty. Ltd., Acton, Australia).24e27 This method analyzes both eyes simultaneously, and although it is promising, it cannot differentiate between the rod and cone systems. Retinitis pigmentosa encompasses a group of progressive retinal degeneration diseases that predominantly affect the rod photoreceptor system, resulting in night blindness in the early phase of the disease and loss of peripheral vision that progresses to tunnel vision. In later stages of RP, degeneration of cone photoreceptors causes progressive decline of visual acuity.28 Disease progression is monitored by electroretinography and perimetry.3,29,30 However, poor testeretest repeatability in patients with RP, specifically in areas with VF deficits,5,6,30,31 limits the ability to assess disease progression and particularly to design and interpret clinical trials of potential therapeutic agents. We recently demonstrated a proof of concept for using a chromatic multifocal pupillometer for detection of VF defects in RP patients. The pupil responses of healthy volunteers and RP patients were recorded at 13 different locations of the 30 VF in response to blue- and red-light stimuli (peak, 485 and 640 nm, respectively; light intensity, 40 cd/m2; target size, 64 mm2).32 Retinitis pigmentosa patients demonstrated a significantly reduced percentage of pupil contraction (PPC) compared with healthy participants in testing conditions that emphasized rod contribution (blue light) in nearly all VF locations. By contrast, the PPC in responses to red light (which emphasizes cone contribution) was reduced significantly in RP patients compared with healthy participants, mostly in peripheral locations. In central locations, there was no significant difference between the PPC of RP patients and healthy participants in response to red light. In a second study, we demonstrated that RP patients demonstrated significantly lower PPC in response to blue light in peripheral locations of the central VF than healthy participants.33 Furthermore, in both studies, minimal PPC was recorded in RP patients in areas that were not detected in dark-adapted chromatic Goldmann perimetry.32,33 In these studies, we evaluated only a single parameter of the pupil response, the PPC. These studies suggested that VF defects as well as rod and cone function may be assessed in RP patients using a chromatic multifocal pupillometer. In the current study, we examined the dynamics of the pupil response in the central VF of RP patients and healthy participants. We analyzed additional parameters of the pupil light response, the maximal contraction velocity (MCV), and the latency of MCV (LMCV) to determine the effect of retinal degeneration on these parameters. To the best of our knowledge, this is the first report of evaluation of the LMCV parameter in pupillometry studies. The central VF of RP patients was assessed using a chromatic multifocal pupillometer and was compared with the patients’ chromatic dark-adapted Goldmann VF (CDAGVF)2 results as well as the pupillometry results of healthy participants. We report that RP patients demonstrated significantly lower PPC and MCV in areas that were reported as nonseeing by CDA-GVF and that the mean
2
absolute deviation in the LMCV parameter between different test point locations was significantly higher in RP patients and may present a valuable parameter as a diagnostic tool for RP.
Methods Participants The Sheba Medical Center Institutional Review Board Ethics Committee approved this trial. The study was conducted according to the tenets of the Declaration of Helsinki and was registered at www.clinicaltrials.gov (identifier, NCT02014389). Informed written consent was obtained from all participants. Sixteen healthy volunteers, age matched with patients (see below; 6 men and 10 women; mean age standard deviation, 38.415.6 years; range, 26e77 years) were included in the study. Inclusion criteria were normal eye examination results, best-corrected visual acuity of 20/20, normal color vision, no history of or current ocular disease, no use of any topical or systemic medications that could adversely influence efferent pupil movements, and normal 24-2 Swedish interactive threshold algorithm results, developed for the Humphrey standard perimeter (Humphrey Field Analyser II, Swedish interactive threshold algorithm 24-2; Carl Zeiss Meditec, Inc., Jena, Germany). The study patient group comprised 13 patients with RP (3 women and 10 men; mean age standard deviation, 36.1514.6 years; range, 20e65 years). Inclusion criteria for RP patients were typical abnormal fundus appearance and previously recorded electroretinography results that were abnormal under scotopic or photopic conditions or both (in compliance with the protocol of the International Society for Clinical Electrophysiology of Vision)34 and typical abnormal kinetic chromatic Goldmann test results (loss of VF that is either concentric or that began superiorly and subsequently demonstrated an arcuate scotoma that progressed either from the nasal or the temporal side; incomplete midperipheral ring scotoma that broke through into the periphery; or a residual central VF, with blue and red isopters that were either superimposed or in which the isopter in response to the red stimulus was larger than that in response to the blue stimulus2,3). Exclusion criteria were a concurrent ocular disease and any other condition affecting the pupil response to light. Data recorded for all patients included gender, diagnosis, and electroretinography responses. Patients were tested for best-corrected visual acuity and for color vision by the Farnsworth D15 test. The right eyes of both healthy and RP participants were examined.
Light Stimuli Light stimuli were presented using a Ganzfeld dome apparatus (Accutome, Inc, Malvern, PA; Fig 1) placed 330 mm from the patient’s eye. All tests were performed in a dark room. The untested eye was covered. Participants were asked to fixate on a white-light fixator (0.9 cd/m2; Fig 1B, white arrow) at the center of the dome. Stimuli were presented from 76 targets (lightemitting diodes) with diameter of 1.8 mm2 in a VF of 16.2 . The wavelength and intensity of light stimuli selected for this study were 6255 nm and 1000 cd/m2 for long-wavelength stimuli (red light) and 4855 nm and 200 cd/m2 for short-wavelength stimuli (blue light). The light intensities were chosen after preliminary calibrations that enabled us to identify the minimal stimulus intensity that yielded a substantial pupil response in 5 healthy participants. Background luminance was 0.05 cd/m2. Light intensities were determined by measurement with the LS-100
Chibel et al
Chromatic Multifocal Pupillometry
Figure 1. Photographs of the chromatic multifocal pupillometer. A, Side view of the device. B, Front view of the device. The black arrow points to the infrared camera pin hole, and the white arrow points to the fixator light-emitting diode. C, The user interface for the chromatic multifocal pupillometer. D, Pupillometer-measured parameters.
luminance meter (Konica Minolta, Tokyo, Japan). Stimulus duration was 1 second and the interstimulus interval was 4 seconds.
Pupil Measurement Pupil diameter was recorded in real time by a computerized infrared high-resolution camera (the camera pinhole is marked with a black arrow in Fig 1B) that recorded the pupil diameter at a frequency of 30 Hz. The software (Accutome, Inc.) searched for the pupil in every image. Participants were dark adapted for 3 minutes before testing. Tests in which the subject blinked during the first 2.5 seconds after stimulus onset were excluded automatically, and the targets were retested.
Analysis of Pupil Responses During the recording of the pupil diameter in response to each light stimulus, 5 parameters were calculated within the software using the change in pupil diameter over time: the initial diameter of the pupil (in pixels), the minimum pupil diameter (in pixels), PPC, MCV (in pixels per second), and LMCV (in seconds). The PPC was determined using the following formula, as described previously32: PPC ¼ ðInitial Pupil Diameter Minimum Pupil DiameterÞ=ðInitial Pupil DiameterÞ 100 The MCV was determined by calculating the maximum rate at which the pupil contracted from the light stimulus between the initial pupil diameter measurement and the minimum pupil diameter measurement. The LMCV was determined by calculating the time point for each pupil response at which the maximum rate of pupil contraction (MCV) occurred from each light stimulus.
Chromatic Dark-Adapted Visual Field Retinitis pigmentosa patients were tested for kinetic VF by CDAGVF.2 Briefly, a Goldmann perimeter (940-ST; Haag-Streit AG, Liebefeld, Switzerland) was used to map patients’ conventional and 2-color dark-adapted VFs. Patients were dark adapted for 30 minutes before testing. The setting used for stimuli were II3c for the long-wavelength stimulus and 2 log units lower in luminance (II3 eC) for the short-wavelength stimulus. For quantification of functional CDA-GVF, a schematic representation of the pupillometer target points was overlaid on the CDA-GVF test record, and pupillometer targets that were within seeing areas by the CDAGVF were scored as 1. Pupillometer targets that were in nonseeing areas by the CDA-GVF were scored as 0. The fraction of functioning VF was calculated as the sum of scores divided by 76.
Statistical Analysis Statistical analyses were performed using Excel (Microsoft, Redmond, WA) and R software version 3.0.1 (The Free Software Foundation [FSF]). Results were presented as mean standard error (SE). Student t test was used to evaluate demographic differences between patients and controls. Testeretest reliability of the pupil response measurements was calculated using Pearson correlation, and the correlation between pupillometer recordings and CDA-GVF testing were calculated with Spearman r test. Variability in LMCV recordings was measured by the mean absolute deviation, and this measure was compared between the study group and the control group using a 2-sided ManneWhitneyeWilcoxon test. The effects of using fewer test points and the robustness of the LMCV for discrimination was examined via simulation by randomly selecting n test points (n ¼ 5, 10, 15, . 75) and calculating the area under the receiver operating characteristic curve (AUC) obtained using the LMCV
3
Ophthalmology Volume -, Number -, Month 2016
Figure 2. Visual field results showing the mean (A, B) percentage of change of pupil size (PPC), (C, D) maximum contraction velocity (MCV), and (E, F) latency of maximum contraction velocity (LMCV) parameters of healthy participants in response to (A, C, E) red and (B, D, F) blue light at each test target location, indicted by number. Color coding for each parameter is shown on the right. deg ¼ degree.
based on these n points. This was done for 200 repeats, and the mean AUC for each n thus was obtained.
Results Characterization of Pupil Responses of Healthy Participants First we characterized the pupil response parameters in each test point for red- and blue-light stimuli in control participants. Figure 2 demonstrates color-coded maps of mean PPC (Fig 2A, B), MCV (Fig 2C, D), and LMCV (Fig 2E, F) recorded from control participants in each test point location in response to red-light stimuli (Fig 2A, C, E) and in response to blue-light stimuli (Fig 2B, D, F). The mean PPC recorded in response to blue-light stimuli ranged from 13% to 28% at different test point locations
4
(mean SE, 19.40.22%) and was significantly higher compared with the red-light stimuli (range, 7%e22%; mean SE, 120.2%; P < 0.0001, t test), even though the red-light stimuli were presented at a 5-fold higher intensity than the blue-light stimuli. Similarly, the mean MCV was significantly higher in response to blue-light stimuli (range, 28e43 pixels/second; mean SE, 370.3 pixels/second) compared with the red-light stimuli (range, 18e36 pixels/second; mean SE, 23.173 pixels/second; P ¼ 2107, t test). There was no significant difference in the mean LMCV between the red and blue light (range, 0.6e0.8 seconds; P ¼ 0.11, t test). Higher PPC was recorded in central locations of the VF compared with peripheral locations in response to both red and blue light (Fig 2A, B). A similar pattern of faster mean MCV in central locations compared with peripheral locations was demonstrated clearly in response to red-light stimuli (Fig 2C). The centraleperipheral gradient pattern was less evident in
Chibel et al
Chromatic Multifocal Pupillometry
Figure 3. Scatterplots showing the test versus retest of (A, B) percentage of change of pupil size (PPC), (C, D) maximum contraction velocity (MCV), and (E, F) latency of maximum contraction velocity (LMCV) parameters in serial testing of 6 healthy participants in response to (A, C, E) blue and (B, D, F) red light. sec ¼ second.
response to blue light (Fig 2D). The LMCV parameter was relatively constant throughout the VF field in response to both red and blue light, with an interquartile range of 0.6 to 0.7 seconds. The longest LMCV values were measured at peripheral test points in response to both wavelengths (Fig 2E, F). Test reliability was assessed by retesting 6 healthy controls. Good linear correlation between the test and the retest was demonstrated for the parameter PPC for both colors (blue: R2 ¼ 0.762; P < 0.0001; red: R2 ¼ 0.721; P < 0.0001; Fig 3A, B). The parameter MCV demonstrated lower but still reasonable correlation between test and retest (blue: R2 ¼ 0.513; P < 0.0001; red: R2 ¼ 0.522; P < 0.0001; Fig 3C, D). The lowest correlation between test and retest was recorded for the LMCV parameter (blue: R2 ¼ 0.419; P < 0.0001; red: R2 ¼ 0.208; P < 0.0001; Fig 3E, F).
Retinitis Pigmentosa Patients Demonstrated Diminished Pupil Responses in Correlation with Disease Severity Thirteen RP patients were divided into 2 groups based on their CDA-GVF testing results. Group A consisted of 5 patients with
some functional CDA-GVF in response to red and blue light, whereas group B included 8 patients with a severe disease with no detection of either blue or red light by CDA-GVF (Table 1). Figure 4 demonstrates grey-scale maps of mean PPC (Fig 4AeC), MCV (Fig 4DeF), and LMCV (Fig 4GeI) recorded in each test point location in response to blue light in healthy participants (Fig 4A, D, G) and in patients with intermediate retinal degeneration (group A; Fig 4B, E, H) and severe retinal degeneration (group B; Fig 4C, F, I). Color coding was set to resemble the Humphrey perimeter’s output, with a white color for normal values and darker colors for values that were less than normal. Normal values were set as the mean of healthy participants in each test point location (Figs 2 and 4). Deviation from normal values was determined based on the SEs calculated for each parameter in each target point in the healthy participants. Thus, for PPC and MCV, the darkest color was used for test points in which the mean of patients was less than 5 SEs from the mean of healthy participants in those points. For LMCV, the darkest color was used for test points in which the mean of patients was more than 5 SEs from the mean of healthy participants in those points. In group B, mean PPC and MCV in response to the blue light were less than 5 SEs from the mean of healthy participants in nearly all test point locations (Fig 4C, F).
5
ND
F ¼ female; M ¼ male; ND ¼ not detected; RP ¼ retinitis pigmentosa; US ¼ Usher syndrome. *Group A ¼ some light detection in response to both red and blue by chromatic dark-adapted Goldmann visual field; group B ¼ no light detection in response to at least 1 color by chromatic dark-adapted Goldmann visual field.
ND 33 23 30 26 46 ND ND 38 26 ND ND ND ND 26 33 5 5 13 ND ND 22 10 ND ND ND ND 16 6 ND 31 16 ND ND 26 17 ND ND ND ND 16 21 ND 9 7 ND ND 9 10 ND ND ND ND 39 42 36 51 44 ND ND 57 44 ND ND ND ND 24.5 17 ND 30 19 ND ND 14 17 ND ND ND ND 95 15 ND 16 15 ND ND 6 15 ND
ND 100 ND ND 78 ND ND ND ND ND ND ND ND ND 74 ND ND 49 ND ND ND ND ND ND ND ND B A A A A A B B B B B B B 20/40 20/3600 20/25 20/32 20/3600 20/20 20/50 20/63 20/64 20/640 20/20 20/30 20/50 RP RP RP RP US RP RP RP RP RP RP RP RP 34 44 34 32 20 31 27 55 21 22 28 65 58 M M F F M F M M M M M M M 1 2 3 4 5 6 7 8 24 25 26 29 30
Group* Gender
Age (yrs)
Diagnosis
Best-Corrected Visual Acuity Patient No.
6
ND 149 11 5 34 7 ND ND 26 17 ND ND ND
Implicit Time (ms) Maximum (mm) Implicit Time (ms)
Single-Flash A-Wave
Maximum (mm) Implicit Time (ms)
Dark-Adapted B-Wave
Maximum (mm) Implicit Time (ms)
Dark-Adapted A-Wave
Maximum (mm) Implicit Time (ms)
Isolated Rod Response
Electroretinography Results
Table 1. Summary of Patient Characteristics
Maximum (mm)
Single-Flash B-Wave
Ophthalmology Volume -, Number -, Month 2016 Similarly, the mean LMCV was more than 5 SEs from the mean of healthy participants in 68 of 76 test points (89% of the VF; Fig 4I). By contrast, the mean PPC recorded in patients in group A was equal to or was lower by less than 2 SEs from the mean of healthy participants in most of the VF test points (50 test points, 66% of the VF; Fig 4B). In 26 test points (34% of the VF) that were located mostly in the periphery of the VF, the mean PPC of group B patients was less than 2 SEs from the mean of healthy participants. The MCV and LMCV parameters also demonstrated an intermediate defect in pupil response. Thus, in 33 test point locations (43% of the VF), the mean MCV was less than 3 SEs from the mean of healthy participants (Fig 4E), and in 25 test point locations (33% of the VF), the LMCV was less than 3 SEs from the mean of healthy participants (Fig 4E, H). Figure 5 demonstrates the pupil responses to red light in healthy participants and RP patients. Retinitis pigmentosa patients from both groups demonstrated lower PPC and MCV and longer LMCV compared with healthy participants, but to a smaller extent than the response to the blue light. Thus, in group B, the mean PPC and MCV were less than 5 SEs from the mean of healthy participants in 35 (46% of the VF) and 57 (75% of the VF) test points, respectively (Fig 5C, F). Similarly, the mean LMCV was more than 5 SEs from the mean of healthy participants in 50 test points (66% of VF) compared with 68 points (89% of the VF) in response to blue light (Fig 5I). Group A demonstrated a milder decline in pupil responses to red light compared with group B, with only 4 (5% of the VF) and 9 (12% of the VF) test points in which the PPC and MCV were less than 5 SEs from the mean of healthy participants, respectively (Fig 5B, E). The mean LMCV was more than 5 SEs from the mean of healthy participants in 35 test points (46% of the VF; Fig 5H).
Variability in Latency of Maximum Contraction Velocity as a Diagnostic Tool for Retinitis Pigmentosa As shown in Figures 2 and 4, LMCV was relatively constant in healthy participants in response to blue and red light in most test point locations, ranging from 0.6 to 0.8 seconds. By contrast, this parameter was highly variable between different test point locations in RP patients, ranging from 0.6 to 1.7 seconds and from 0.6 to 1.4 seconds in response to the blue and red light, respectively (Figs 4 and 5). To evaluate the extent of the variability in LMCV between different test point locations of the VF, the mean response for each subject was determined (i.e., the mean LMCV among the subject’s 76 test points). Then, the mean absolute deviation was calculated as the mean of the absolute difference between the mean and the measurements in each of the test points. Figure 6A demonstrates for each participant a boxplot depicting the distribution of the LMCV parameter for all test points in response to the red light. The mean absolute deviation in LMCV in response to the red light between different test points of each participant was significantly higher in patients with RP than in healthy participants (P < 106, ManneWhitneyeWilcoxon test). In addition, the ManneWhitneyeWilcoxon statistic test indicated that a classification method based on measurement of LMCV in response to red light would have an AUC of 0.97. Similarly, the mean absolute deviation of LMCV in response to the blue light between different test points of each participant was significantly higher in patients with RP than in healthy participants (P ¼ 104, WilcoxoneManneWhitney test; AUC, 0.93; data not shown). There was no significant difference in the absolute mean
Chibel et al
Chromatic Multifocal Pupillometry
Figure 4. Visual field (VF) results showing the mean (AeC) percentage of change of pupil size (PPC), (DeF) maximum contraction velocity (MCV), and (GeI) latency of maximum contraction velocity (LMCV) parameters recorded in response to blue light in (A, D, G) healthy participants and in patients from (B, E, H) group A and (C, F, I) group B in each of the 76 test targets of the 16.2 VF. Color coding is based on number of standard errors (SEs) from the mean of healthy participants and is shown on the right. White ¼ mean value equals or is less than 1 SE from the mean of healthy participants in each location; dark gray ¼ mean values differs in more than 5 SEs from the mean of healthy participants in each location. deg ¼ degree.
deviation of the PPC and MCV parameters between RP patients and healthy controls (data not shown). The mean absolute deviation in LMCV in response to the red light correlated negatively with the fraction of functional subjective VF as determined by CDA-GVF (Spearman r, 0.45; P ¼ 0.13; Fig 6B). The highest mean absolute deviation in LMCV (>0.3) was found in patients with no light detection by CDA-GVF. By contrast, lower mean absolute deviation in LMCV (<0.3) was demonstrated in patients with some functional CDA-GVF.
Clustering Analysis Next we examined whether fewer test point locations could be used for RP diagnosis in an attempt to optimize the analysis of pupillometer-based perimetry and reduce testing time. Figure 7 shows the mean and standard deviation of the AUC obtained after a random selection of test locations. Randomly reducing the number of test points to a total of 15 points did not significantly reduce the AUC in RP diagnosis based on the absolute mean deviation of LMCV. The AUC of 0.90 (highlighted by a dashed
red line) suggests that the probability of LMCV to discriminate successfully a (randomly selected) RP patient from a (randomly selected) healthy subject is 0.9.
Individual Retinitis Pigmentosa Cases To illustrate the pattern of recorded pupil response values compared with the results of subjective CDA-GVF testing, the individual reports for 3 RP patients are presented in Figures 8, 9, and 10. Patient 30: No Light Detection by Chromatic Dark-Adapted Goldmann Visual Field. Patient 30 showed no light detection by CDA-GVF (Fig 8A, E). The PPC and MCV parameters were less than 5 SEs from the mean of healthy participants in 75 of the 76 test points (99% of the VF) in response to blue light (Fig 8B, C). The LMCV parameter was more than 5 SEs from the mean of healthy participants in 48 of the 76 test points (63% of the VF; Fig 8D). The pupil responses to red light also were diminished significantly throughout the VF, with 69 and 74 of the 76 test points (91%
7
Ophthalmology Volume -, Number -, Month 2016
Figure 5. Visual field (VF) results showing the mean (AeC) percentage of change of pupil size (PPC), (DeF) maximum contraction velocity (MCV), and (GeI) latency of maximum contraction velocity (LMCV) parameters in response to red light in (A, D, G) healthy participants and in patients from (B, E, H) group A and (C, F, I) group B in each of the 76 test targets of the 16.2 VF. Color coding is based on number of standard errors (SEs) from the mean of healthy participants and is shown on the right, as described in Figure 4. deg ¼ degree.
and 77% of the VF), respectively, presenting PPC and MCV less than 4 SEs from the mean of healthy participants (Fig 8F, G). The LMCV parameter was more than 5 SEs from the mean of healthy participants in 38 test points (50% of the VF; Fig 8H). The mean absolute deviation in LMCV in response to the red light for this patient was the largest recorded in this study (0.47 second; Fig 6A). Patient 4: Tunnel Vision by Chromatic Dark-Adapted Goldmann Visual Field. Patient 4 demonstrated tunnel vision on CDA-GVF (Fig 9A, E). The map of the PPC parameter in response to blue light correlated well with the CDA-GVF map. Thus, in peripheral test point locations (nonseeing by CDA-GVF), the PPC values were less than 5 SEs from the mean of healthy participants, whereas in central locations, PPC values were only 1 to 3 SEs less than the mean of healthy participants (Fig 9B). The map of MCV demonstrated substantial reduction of MCV throughout the VF, with MCV less than 4 SEs from the mean of healthy participants in all 76 test points (100% of the VF; Fig 9C). The LMCV parameter was 1 to 2 SEs from the mean of healthy participants in 20 test points (26% of the VF), mostly in central locations (Fig 9D).
8
A similar tunnel vision pattern of pupil responses was obtained in response to the red light. The PPC and MCV parameters were equal to or only 1 SE from the mean of healthy participants in 14 and 8 central test point locations, correlating to 18% and 11% of the VF, respectively. By contrast, the PPC and MCV recorded in nearly all peripheral test points were more than 5 SEs less than the mean of healthy participants (Fig 9F, G). The LMCV parameter was close to normal (1e2 SEs from the mean of healthy participants) in 38 test points (50% of the VF). Most of these test points were located at the center of the VF (Fig 9H). The mean absolute deviation in LMCV in response to the red light for this patient was intermediate (0.26 second; Fig 6A). Patient 3: Larger Fraction of Functional Chromatic DarkAdapted Goldmann Visual Field. Patient 3 had a significant functional portion of the 16.2 VF as determined by the subjective CDA-GVF (Fig 10A, E). The pupil responses of this patient to both the red and blue light were equal to or 1 to 2 SEs from the mean of healthy participants in most test points located in functional areas determined by the CDA-GVF. Most areas that demonstrated substantially reduced PPC and MCV and longer LMCV compared with control values (more than 2 SEs from the mean of healthy
Chibel et al
Chromatic Multifocal Pupillometry
Figure 6. The variability in parameter latency of maximum contraction velocity (LMCV) in response to red-light stimuli between different visual field (VF) test point locations in retinitis pigmentosa (RP) patients correlates with their chromatic dark-adapted Goldmann VF (CDA-GVF) results. A, Box-and-whisker plot showing that RP patients demonstrate significantly higher variability in LMCV in response to red-light stimuli between different VF locations compared with control subjects. The absolute difference between the mean LMCV (of the 76 points) and the recordings in 76 points was determined for each subject (y-axis). x-axis ¼ subject number. Boxes represent the first and third quartiles and the band inside the box represents the second quartile (the median). The whiskers extend to the most extreme data point, which is no more than 1.5 times the interquartile range from the box. B, Graph showing negative linear correlation between the mean absolute deviation in LMCV in response to red light and CDA-GVF. y-axis ¼ the mean absolute deviation of LMCV (of the 76 points) in response to red light. x-axis ¼ the fraction of functioning VF in response to red light that was calculated for RP patients based on their subjective CDA-GVF results. Solid line ¼ linear regression.
participants) were located mostly in nonseeing areas by CDA-GVF (Fig 10). The mean absolute deviation in LMCV in response to the red light for this patient was small (0.16 second; Fig 6A).
Discussion This study demonstrated the feasibility of using a chromatic multifocal pupillometer for objective assessment of VF
Figure 7. The computational random selection analysis of the mean absolute deviation in latency of maximum contraction velocity parameter: the mean and standard deviation of the area under the receiver operating characteristic curve (AUC) obtained after a random selection of test locations (y-axis) versus the number of test points used (x-axis).
defects in patients with RP. Test point locations in which PPC and MCV were less than 4 SEs from the mean of healthy participants correlated with areas that were abnormal (nonseeing) by CDA-GVF. The RP patients with severe VF loss presented more testing points that differed substantially from the mean of healthy participants compared with patients with a moderate loss of VF, particularly in response to the blue-light stimuli. Patients with some functional VF demonstrated reduced pupil light response (PLR) particularly in peripheral test points and in response to the blue-light stimuli. The pathologic features of RP are characterized by a loss of rod function that exceeds the reduction of cone function, and the VF loss typically begins with peripheral VF constriction.3,28 Our finding that pupil response to blue light in RP patients was more affected than the pupil response to red-light stimuli strongly suggests that the pupil response to blue light measured by our chromatic pupillometer in response to focal light stimuli is mediated mainly by rods, whereas the pupil response to red light is mediated mainly by cones. Hence, a chromatic multifocal pupillometer may enable objective noninvasive assessment of the function of rods and cones at distinct locations of the VF for the first time. These results are supported by our previous study that demonstrated reduced pupil responses in peripheral VF locations in response to blue light32 and by the studies from Kawasaki et al,35 Kardon et al,36,37 and Lorenz et al38 that demonstrated that transient pupil response to full-field low-intensity blue-light stimulus reflects rod activity and that transient pupil response to red light is driven predominantly by cones. Curcio et al39 reported that the average horizontal diameter of the rod-free zone in the human fovea is 0.350 mm (1.25 ).
9
Ophthalmology Volume -, Number -, Month 2016
Figure 8. A, E, Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in patient 30 demonstrated no light detection. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of maximum contraction velocity (LMCV) parameters of this patient in response to (BeD) blue-light and (FeH) red-light stimuli recorded in each of the 76 points of the 16.2 visual field are presented. Color coding is the same as described in Figure 4. deg ¼ degree.
Because the 4 most central pupillometer test point locations are located at 1.85 , the pupil responses to blue light recorded from these 4 central test points may be mediated by foveal rods. Since the pupil responses recorded in this study were transient and because dim light was used in the current test protocol, it is less likely that intrinsically photosensitive retinal ganglion cells are involved in pupil response to blue-light stimuli presented at the 4 central test points. However, because cones are highly concentrated in the fovea, it is possible that some of the pupil response to blue light in the 4 central test points is mediated by S cones. We plan to examine this possibility in a future study by testing coneerod dystrophy patients. The intensity of blue-light stimulus used in our study was 5-fold lower than that of the red-light stimulus. Nevertheless, pupil responses to blue-light stimulus in healthy participants were stronger than the responses to red-light stimuli in the same test point locations (Fig 2). These findings correlate with those of our previous study using the first chromatic pupillometer prototype32 and with the findings of Kawasaki et al,35 Kardon et al,36,37 and Lorenz et al,38 and may be explained by the lower number of cones compared with rods in the human retina39 and by the smaller
10
receptive fields of cones and their lower sensitivity for light compared with rods. The LMCV parameter recorded in response to red-light stimuli seems to be a useful tool for noninvasive and objective diagnosis of RP with an AUC of 0.97. Our computer-based random clustering analysis suggested that shortening test duration may be possible with computational clustering, without reducing the sensitivity and accuracy of RP diagnosis. Thus, testing only 15 test points in response to red-light stimuli, which is predicted to take 1 minute, would enable diagnosis of RP with an AUC of 0.9. Importantly, our study group included RP patients at different stages of the diseases, some with substantial functional 16.2 VF (such as patient 3) and some patients who had no light detection (such as patient 30), further emphasizing the high specificity and sensitivity of the LMCV score. In coming clinical trials using a chromatic multifocal pupillometer, we will test a larger VF (30 ) using a larger cohort of patients and healthy participants. The future study also will include more thorough computational optimization for clustering based on the promising results of random reduction of target locations. In addition, future trials are planned in which the pupillometry testing of patients at early stages and healthy participants will be
Chibel et al
Chromatic Multifocal Pupillometry
Figure 9. A, E, Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in patient 3 demonstrating the subjective visual field (VF) for (A) blue and (E) red light. The dashed green line marks the borders of the 16.2 VF. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of MCV (LMCV) parameters of the patient in response to (BeD) blue-light and (FeH) red-light stimuli recorded in each of the 76 points of the 16.2 VF are presented. Color coding is the same as described in Figure 4. deg ¼ degree.
performed in a masked fashion to examine the applicability of pupillometry testing as a diagnostic tool. Previous studies suggested direct correlation between pupil response amplitude, maximum velocity, and latency in response to full-field white-light stimulus. The authors noted that most of the latency in healthy participants was the result of delay in iris smooth muscle contraction and that a relatively small part is the result of conduction along the pupil reflex pathway.40 Because apoptosis in the iris muscle increases with age, we included agematched groups with a wide range of ages (26e77 years). To the best of our knowledge, our study enables mapping different pupil light response parameters in different locations of the VF in response to red and blue light for the first time. Our findings suggest that in most cases, maximum pupil contraction velocity correlated directly with change in pupil diameter. However, some exceptions were noted (for example, in patient 4). A future clinical trial with a larger study group may enable determination of the correlation between the different pupil response parameters in this new setting of response to focal chromatic light stimuli. One of the major limitations of subjective perimetry is testeretest variability. A good correlation was demonstrated in PPC and MCV in testeretest analysis. The large bin size
of the LMCV parameter (0.033 second) and outliers may be the cause for the low testeretest correlation of LMCV. A future clinical trial will examine the test reliability in repeated testing of a larger number of healthy participants and patients. The chromatic multifocal pupillometer device presented here was built to enable mapping of the central VF. We chose to examine the device in RP patients because of their pathologic features that facilitate the differentiation between cone and rod function. Future clinical trials will include patients with other blinding diseases, such as patients with macular degeneration and glaucoma, using a device that enables 30 VF testing. The major limitation of this study is the size of the study cohorts. This is a preliminary study that is being followed by a larger clinical trial with more types of diseases. Several studies demonstrated a correlation between structural abnormalities on optical coherence tomography (OCT), including decreased outer nuclear layer and outer segment thickness and the loss of local VF sensitivity of RP patients.41e44 Wen et al45 demonstrated strong positive correlation between cone function preservation measured by multifocal electroretinography, VF sensitivity, and the remaining thickness of the photoreceptor layer by spectral-domain OCT in RP patients. Recent adaptive
11
Ophthalmology Volume -, Number -, Month 2016
Figure 10. Chromatic dark-adapted Goldmann visual field (CDA-GVF) results in retinitis pigmentosa patient 4 demonstrating the subjective visual field (VF) for (A) blue and (E) red light. The dashed green line marks the borders of the 16.2 VF. The (B, F) percentage of change of pupil size (PPC), (C, G) maximum contraction velocity (MCV), and (D, H) latency of MCV (LMCV) parameters of this patient in response to (BeD) blue-light and (FeH) redlight stimuli recorded in each of the 76 points of the 16.2 VF are presented. Color coding is the same as in Figure 4. deg ¼ degree.
optics scanning laser ophthalmoscopy studies demonstrated early detection of reduced cone density and irregular cone mosaic spatial arrangement in the macula of RP patients with best-corrected visual acuity of 20/20.46e48 Future trials will include follow-up pupillometry and spectral-domain OCT imaging that will allow analysis of longitudinal changes in pupillary response with respect to disease progression and examination of the correlation between functional pupillometry findings and spectraldomain OCT structural findings. In addition, it would be interesting to examine the correlation between the chromatic multifocal pupillometry function testing and adaptive optics scanning laser ophthalmoscopy imaging. In the current study, the pupillometer test results were compared with CDA-GVF that enables evaluation of cone and rod responses. In future studies, the pupillometer results will be compared with the more commonly used Humphrey perimetry (Carl Zeiss Meditec, Inc., Jena, Germany). Taken together, our results suggest direct correlation between seeing and pupil response function of the retina. Different parameters of pupil response can be used for VF determination (PPC and MCV) and disease diagnosis (stimulus color and LMCV). The pupillometer test is predicted to be less stressful for tested subjects because patients are unaware of the test results and the test may facilitate
12
objective perimetry and assessment of function of retinal photoreceptors with minimal patient cooperation and minimal technician training.
References 1. Sharma P, Sample PA, Zangwill LM, et al. Diagnostic tools for glaucoma detection and management. Surv Ophthalmol 2008;53(suppl 1):S17–32. 2. Rotenstreich Y, Fishman GA, Lindeman M, et al. The application of chromatic dark-adapted kinetic perimetry to retinal diseases. Ophthalmology 2004;111:1222–7. 3. Grover S, Fishman GA, Brown J Jr. Patterns of visual field progression in patients with retinitis pigmentosa. Ophthalmology 1998;105:1069–75. 4. Trope GE, Britton R. A comparison of Goldmann and Humphrey automated perimetry in patients with glaucoma. Br J Ophthalmol 1987;71:489–93. 5. Kim LS, McAnany JJ, Alexander KR, et al. Intersession repeatability of Humphrey perimetry measurements in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2007;48: 4720–4. 6. Seiple W, Clemens CJ, Greenstein VC, et al. Testeretest reliability of the multifocal electroretinogram and Humphrey
Chibel et al
7. 8.
9. 10. 11. 12. 13.
14. 15. 16. 17. 18. 19. 20. 21. 22.
23. 24. 25. 26. 27.
Chromatic Multifocal Pupillometry
visual fields in patients with retinitis pigmentosa. Documenta Ophthalmologica 2004;109:255–72. Heijl A, Lindgren A, Lindgren G. Test-retest variability in glaucomatous visual fields. Am J Ophthalmol 1989;108:130–5. Artes PH, Iwase A, Ohno Y, et al. Properties of perimetric threshold estimates from full threshold, SITA standard, and SITA fast strategies. Invest Ophthalmol Visual Sci 2002;43: 2654–9. Turalba AV, Grosskreutz C. A review of current technology used in evaluating visual function in glaucoma. Semin Ophthalmol 2010;25:309–16. Chauhan BC, Garway-Heath DF, Goni FJ, et al. Practical recommendations for measuring rates of visual field change in glaucoma. Br J Ophthalmol 2008;92:569–73. Advanced Glaucoma Intervention Study Group. Advanced Glaucoma Intervention Study 2: visual field test scoring and reliability. Ophthalmology 1994;101:1445–55. Harms H. Grundlagen, methodik und bedeutung der pupillenperimetrie für die physiologie und pathologie des sehorgans. Graefes Arch Clin Exp Ophthalmol 1949;149:1–68. Reuther R, Krastel H, Alexandridis E. Disturbances of the pupil reflex associated with homonymous hemianopic paracentral scotoma [in German]. Arch Psychiatr Nervenkr 1981;229:259–66. Asakawa K, Shoji N, Ishikawa H, et al. New approach for the glaucoma detection with pupil perimetry. Clin Ophthalmol 2010;4:617–23. Hong S, Narkiewicz J, Kardon RH. Comparison of pupil perimetry and visual perimetry in normal eyes: decibel sensitivity and variability. Invest Ophthalmol Vis Sci 2001;42:957–65. Aoyama T. Pupillographic perimetry. the application to clinical cases [in Japanese]. Nihon Ganka Gakkai Zasshi 1977;81: 1527–38. Chang DS, Xu L, Boland MV, et al. Accuracy of pupil assessment for the detection of glaucoma: a systematic review and meta-analysis. Ophthalmology 2013;120:2217–25. Alexandridis E, Krastel H. Pupillographic perimetry using the “octopus”. Fortschr Ophthalmol 1989;86:692–4. Yoshitomi T, Matsui T, Tanakadate A, et al. Comparison of threshold visual perimetry and objective pupil perimetry in clinical patients. J Neuroophthalmol 1999;19:89–99. Schmid R, Luedtke H, Wilhelm BJ, et al. Pupil campimetry in patients with visual field loss. Eur J Neurol 2005;12:602–8. Skorkovska K, Wilhelm H, Ludtke H, et al. How sensitive is pupil campimetry in hemifield loss? Graefes Arch Clin Exp Ophthalmol 2009;247:947–53. Skorkovska K, Ludtke H, Wilhelm H, et al. Pupil campimetry in patients with retinitis pigmentosa and functional visual field loss. Graefes Arch Clin Exp Ophthalmol 2009;247: 847–53. Jensen W. A description of a method for objective perimetry [in German]. Graefes Arch Clin Exp Ophthalmol 1976;201: 183–91. Sabeti F, Maddess T, Essex RW, et al. Multifocal pupillographic assessment of age-related macular degeneration. Optom Vis Sci 2011;88:1477–85. Maddess T, Ho YL, Wong SS, et al. Multifocal pupillographic perimetry with white and colored stimuli. J Glaucoma 2011;20:336–43. Carle CF, James AC, Kolic M, et al. Blue multifocal pupillographic objective perimetry in GlaucomaBlue mfPOP in glaucoma. Invest Ophthalmol Vis Sci 2015;56:6394–403. Sabeti F, Nolan CJ, James AC, et al. Multifocal pupillography identifies changes in visual sensitivity according to severity of diabetic retinopathy in type 2 DiabetesmfPOP to identify
28. 29. 30. 31. 32.
33.
34. 35.
36.
37. 38.
39. 40. 41. 42. 43.
44.
45.
46.
patients with and without DR. Invest Ophthalmol Vis Sci 2015;56:4504–13. Hartong DT, Berson EL, Dryja TP. Retinitis pigmentosa. Lancet 2006;368:1795–809. Birch DG, Anderson JL, Fish GE. Yearly rates of rod and cone functional loss in retinitis pigmentosa and cone-rod dystrophy. Ophthalmology 1999;106:258–68. Holopigian K, Greenstein V, Seiple W, et al. Rates of change differ among measures of visual function in patients with retinitis pigmentosa. Ophthalmology 1996;103:398–405. Ross DF, Fishman GA, Gilbert LD, et al. Variability of visual field measurements in normal subjects and patients with retinitis pigmentosa. Arch Ophthalmol 1984;102:1004–10. Skaat A, Sher I, Kolker A, et al. Pupillometer-based objective chromatic perimetry in normal eyes and patients with retinal photoreceptor dystrophies. Invest Ophthalmol Vis Sci 2013;54:2761–70. Rotenstreich Y, Chibel R, Yahia SH, et al. The first prototype of chromatic pupillometer for objective perimetry in retinal degeneration patients. SPIE BiOS Int Soc Optics Photonics 2015. 93070Qe93070Q-7. Marmor MF, Fulton AB, Holder GE, et al. ISCEV standard for full-field clinical electroretinography (2008 update). Doc Ophthalmol 2009;118:69–77. Kawasaki A, Crippa SV, Kardon R, et al. Characterization of pupil responses to blue and red light stimuli in autosomal dominant retinitis pigmentosa due to NR2E3 mutation. Invest Ophthalmol Vis Sci 2012;53:5562–9. Kardon R, Anderson SC, Damarjian TG, et al. Chromatic pupil responses: preferential activation of the melanopsinmediated versus outer photoreceptor-mediated pupil light reflex. Ophthalmology 2009;116:1564–73. Kardon R, Anderson SC, Damarjian TG, et al. Chromatic pupillometry in patients with retinitis pigmentosa. Ophthalmology 2011;118:376–81. Lorenz B, Strohmayr E, Zahn S, et al. Chromatic pupillometry dissects function of the three different light-sensitive retinal cell populations in RPE65 deficiency. Invest Ophthalmol Vis Sci 2012;53:5641–52. Curcio CA, Sloan KR, Kalina RE, et al. Human photoreceptor topography. J Comp Neurol 1990;292:497–523. Ellis CJ. The pupillary light reflex in normal subjects. Br J Ophthalmol 1981;65:754–9. Rangaswamy NV, Patel HM, Locke KG, et al. A comparison of visual field sensitivity to photoreceptor thickness in retinitis pigmentosa. Invest Ophthalmol Vis Sci 2010;51:4213–9. Hood DC, Lazow MA, Locke KG, et al. The transition zone between healthy and diseased retina in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2011;52:101–8. Jacobson SG, Cideciyan AV, Aleman TS, et al. Usher syndromes due to MY07A, PCDH15, USH2A or GPR98 mutations share retinal disease mechanism. Hum Mol Genet 2008;17:2405–15. Jacobson SG, Aleman TS, Sumaroka A, et al. Disease boundaries in the retina of patients with usher syndrome caused by MY07A gene mutations. Invest Ophthalmol Vis Sci 2009;50:1886–94. Wen Y, Klein M, Hood DC, et al. Relationships among multifocal electroretinogram amplitude, visual field sensitivity, and SD-OCT receptor layer thicknesses in patients with retinitis pigmentosa. Invest Ophthalmol Vis Sci 2012;53:833–40. Makiyama Y, Ooto S, Hangai M, et al. Macular cone abnormalities in retinitis pigmentosa with preserved central vision using adaptive optics scanning laser ophthalmoscopy. PLoS One 2013;8:e79447.
13
Ophthalmology Volume -, Number -, Month 2016 47. Park SP, Lee W, Bae EJ, et al. Early structural anomalies observed by high-resolution imaging in two related cases of autosomal-dominant retinitis pigmentosa. Ophthal Surg Lasers Imaging 2014;45:469–73.
48. Park SP, Hong IH, Tsang SH, et al. Cellular imaging demonstrates genetic mosaicism in heterozygous carriers of an X-linked ciliopathy gene. Eur J Hum Genet 2013;21: 1240–8.
Footnotes and Financial Disclosures Originally received: January 11, 2016. Final revision: May 21, 2016. Accepted: May 23, 2016. Available online: ---.
Author Contributions: Conception and design: Rotenstreich, Sher Manuscript no. 2016-92.
1
The Goldschleger Eye Institute, Sheba Medical Center, Tel-Hashomer, Israel.
2
Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
3
Department of Ophthalmology, Wolfson Medical Center, Holon, Israel. 4 Biostatistics Unit, Gertner Institute for Epidemiology and Health Policy Research, Ramat Gan, Israel. 5
Unit of Cardiovascular Epidemiology, Gertner Institute for Epidemiology and Health Policy Research, Ramat Gan, Israel.
Financial Disclosure(s): The author(s) have made the following disclosure(s): Y.R.: Financial support (to institution) e Accutome Inc., Malvern, PA; Patent e Sheba Medical Center, Tel-Hashomer, Israel Supported by Accutome, Inc., Malvern, PA; Accutome, Inc., participated in review of the manuscript but had no role in the design or conduct of this research.
14
Analysis and interpretation: Chibel, Sher, Achiron, Skaat, Berchenko, Oberman, Kalter-Leibovici, Freedman, Rotenstreich Data collection: Chibel, Sher, Ben Ner, Mhajna, Hajyahia Obtained funding: Rotenstreich Overall responsibility: Rotenstreich Abbreviations and Acronyms: AUC ¼ area under the receiver operating characteristic curve; CDAGVF ¼ chromatic dark-adapted Goldmann visual field; LMCV ¼ latency of maximum contraction velocity; MCV ¼ maximum contraction velocity; OCT ¼ optical coherence tomography; PPC ¼ percentage of change of pupil size; RP ¼ retinitis pigmentosa; SE ¼ standard error; VF ¼ visual field. Correspondence: Ygal Rotenstreich, MD, Goldschleger Eye Institute, Sheba Medical Center, Tel Hashomer 52621, Israel. E-mail: Ygal.Rotenstreich@sheba. health.gov.il.