Eye Movements, Strabismus, Amblyopia, and Neuro-Ophthalmology

Transcription

1 Eye Movements, Strabismus, Amblyopia, and Neuro-Ophthalmology Chromatic Pupillometry Dissects Function of the Three Different Light-Sensitive Retinal Cell Populations in RPE65 Deficiency Birgit Lorenz, 1 Elisabeth Strohmayr, 1 Steffen Zahn, 1 Christoph Friedburg, 1 Martin Kramer, 2 Markus Preising, 1 and Knut Stieger 1 PURPOSE. The aim of the study was to objectively characterize the function of rods, cones, and intrinsic photosensitive retinal ganglion cells (iprgcs) in patients with RPE65 mutations by using two published protocols for chromatic pupillometry, and to correlate the data with the clinical phenotype. METHODS. The study group comprised 11 patients with RPE65 mutations, and for control purposes, 32 healthy probands and 2 achromats. A custom-made binocular chromatic pupillometer (Bino I) connected to a ColorDome Ganzfeld stimulator was used to assess changes in pupil diameter in response to red (640 nm) and blue (462 nm) light stimuli. Light intensities, stimulus duration, and background varied depending on the protocol used. Results were compared to the clinical phenotype, that is, visual field (Goldmann perimetry), best corrected visual acuity, and full-field stimulus testing (FST). RESULTS. No significant differences in any of the pupil response parameters were observed in intraday or intervisit variability tests. Pupil responses to rod-weighted stimulation were significantly diminished in all RPE65 patients. Pupil responses to cone-weighted stimuli differed among RPE65 patients and did not always correlate with residual visual field and cone sensitivity loss in FST. Pupil responses to iprgc-weighted answers were slightly but significantly diminished, and the postillumination pupil response was significantly increased. CONCLUSIONS. Chromatic pupillometry represents a highly sensitive and objective test to quantify the function of rods, cones, and iprgcs in patients with RPE65 mutations. (Invest Ophthalmol Vis Sci. 2012;53: ) DOI: / iovs Mutations in RPE65 cause a phenotypically heterogeneous disease in humans referred to as either early-onset severe retinal dystrophy (EOSRD) or Leber congenital amaurosis (LCA) type 2. 1,2 The classification depends on whether legal blindness appears within the first years of life (LCA), or during the second decade of life (EOSRD). The clinical symptoms vary From the 1 Department of Ophthalmology, Faculty of Medicine, and the 2 Small Animal Clinic, Faculty of Veterinary Medicine, Justus- Liebig-University Giessen, Giessen, Germany. Supported by von Behring Röntgen Foundation ( ). Submitted for publication April 5, 2012; revised June 6 and June 29, 2012; accepted July 13, Disclosure: B. Lorenz, None;E. Strohmayr, None;S. Zahn, None; C. Friedburg, None;M. Kramer, None;M. Preising, None; K. Stieger, None Corresponding author: Birgit Lorenz, Department of Ophthalmology, Justus-Liebig-University, Giessen, Germany; birgit.lorenz@uniklinikum-giessen.de. and include severe reduction of the visual field, profound night blindness, absent rod electroretinogram (ERG), eventually early on some residual cone ERG, and preferential loss of blue color sensitivity. 2 5 Lack of fundus autofluorescence despite relatively normal-looking retinae represents the hallmark clinical sign of RPE65 deficiency. 4 The range of phenotypic severity has been correlated with the different functional consequences of specific mutations within the RPE65 gene. 5 It has been long thought that the RPE65 protein is uniquely expressed in the retinal pigment epithelium (RPE), but recent findings indicate a possible expression in red/green cones and Muller cells. 6 8 Located in the RPE, the protein is responsible for the isomerization of all-trans retinol into 11-cis retinol, thus supplying predominantly the rods with functional chromophore. 9,10 An extra supply mechanism for red/green cones would also explain the early pronounced function loss of blue cones only 11 (Lorenz B, Stark R, Wegscheider E, et al., manuscript in preparation). Since more than a decade, adeno-associated virus (AAV) mediated gene therapy approaches for the treatment of RPE65 deficiency have been developed, initially in animal models followed by clinical trials in humans Currently, seven different clinical trials in humans have been started or are already completed at different sites worldwide (provided in the public domain by, However, promising results from dog trials, both in terms of morphologic and functional rescue, could not be transferred to human trials directly. Visual field, light sensitivity, and performance in obstacle parcours in dim light are among the subjective and objective readouts that show significant improvement following treatment. 22,23 On the other hand, restoration of ERG recordings and improvement of central cone function resulting in increased visual acuity have not been reported in humans. The reason for this discrepancy between human and canine trial outcome is likely due to differences in the consequences of specific mutations, that is, missense versus null; in disease progression stage at the time of treatment; and in the complex genetic background of humans compared to animal models. 25 Furthermore, technical limitations also account for the relative lack of objective examination results, which are needed for a thorough investigation of the clinical benefit. 3 Especially, electroretinographic methods are not sufficiently sensitive to detect small changes in retinal activity at the site of injection ,23 Hence, highly sensitive and objective methods for the characterization of minimal visual perception by photosensitive cells of the retina are highly desirable in order to improve clinical readout results. Full-field stimulus testing (FST) and white light stimulus pupillometry have been developed to characterize lowest levels of visual perception. 26,27 While FST is a dark-adapted psychophysical testing method and thus relies on the cooperation of the patient, white light stimulus Investigative Ophthalmology & Visual Science, August 2012, Vol. 53, No. 9 Copyright 2012 The Association for Research in Vision and Ophthalmology, Inc. 5641

2 5642 Lorenz et al. IOVS, August 2012, Vol. 53, No. 9 pupillometry detects the summation response of all lightsensitive cells within the retina through measuring the resulting transient pupil constriction. The pupillary light reflex (PLR) controls the light intensity that reaches the retina by linking the irradiance level to the pupil diameter For a long time, it was thought that the PLR, and human vision in general, rely uniquely on the activation of rod photoreceptors in dim light, and short (S)-, medium (M)-, and long (L)-wavelength sensitive cone photoreceptors in daylight, and that the information is passed on to ganglion cells via bipolar, horizontal, and amacrine cell processing. 31 However, approximately a decade ago, Lucas and colleagues 32 demonstrated that rodless and coneless rodent retinae display a reduced but measurable PLR and normal diurnal activity, which was driven by a thitherto unknown light-sensitive photopigment. Such an alternative pathway had already been suggested earlier. 33,34 Shortly thereafter, it was shown that this parallel, nonrod, noncone photoreceptive pathway arises from a separate population of photoreceptive retinal ganglion cells 35 containing the putative photopigment melanopsin. 36 By signaling gross changes in light intensity, these cells serve the subconscious, nonimageforming functions of circadian photo entrainment and pupil constriction Since its initial discovery, several studies have now provided evidence that melanopsin is indeed the photopigment responsible for the photo response of intrinsically photosensitive retinal ganglion cells (iprgcs) in mice. Because the absorption maximum of melanopsin in rodents has been defined as being around 479 nm and the spectral sensitivity of these ganglion cells differs from that of rods and cones, it seemed to be possible to separate rod, cone, and melanopsin components to the PLR by their spectral sensitivity and reaction kinetics. 32 Dacey and colleagues 45 have shown that a comparable population of retinal ganglion cells also exists in primates, and that these cells project to the pupillary control center in the pretectum. The spectral sensitivity of intrinsic melanopsinmediated phototransduction in primates is broad with a peak at around 483 nm. After medicinal block of rods and cones in primates, blue light with high intensity evokes a sustained pupil constriction after light offset, indicating a significant contribution of iprgcs to the sustained component of the PLR. 46 Based on these discoveries, it became evident that the PLR is driven by rod-, cone-, and ganglion cell mediated activity. As the different receptor cells have different absorption maxima and sensitivities, selective activation of each receptor system separately by stimuli of different wavelengths and different intensities is possible. This paradigm of chromatic pupillometry has been explored and optimized by Kardon and colleagues, who developed a protocol to stimulate the different light-sensitive cell populations by using different wavelengths and intensities. 47,48 Other groups have developed protocols for the characterization of iprgc function in glaucoma 49,50 or diabetic retinopathy. 51 All protocols are based on the assumption that rods (absorption maximum at 498 nm) respond to blue light at low luminance values (normal threshold at 3 to 5 log cd/m 2 ) and iprgcs sensitize blue light at much higher luminance levels in order to generate a pupil response. 46 S-cones (absorption maximum at 460 nm) react to luminance values approximately 3 logs higher than rods for eliciting a pupil response. Finally, L/M cones (absorption maximum around 560 nm) can be uniquely stimulated at wavelengths beyond 620 nm, while rods, iprgcs, and S-cones are thought to be insensitive to light stimuli at this wavelength. 52 In this article, we present data for a cohort of 11 patients with RPE65 mutations who underwent testing with both protocols for chromatic pupillometry in order to dissect the residual function of rods, cones, and iprgcs in this phenotypically heterogeneous retinal disorder. We showed that rodmediated pupil constriction was significantly reduced among patients with RPE65 mutations, which correlated well with the clinical phenotype (i.e., best corrected visual acuity [BCVA], visual field, and FST). The status of residual cone function is more complex and does not always correspond to the clinical phenotype of the patients. The function of iprgc is reduced and postillumination pupil response (PIPR) is prolonged. Replicate testing revealed no significant difference for any of the pupillary response parameters. These data proved that chromatic pupillometry is a highly sensitive measurement method and should be added to the panel of examination tools for the characterization of the natural history of retinal degenerative disorders and for quantifying the clinical benefit in patients following experimental treatment protocols. METHODS Subjects Both eyes of 32 healthy subjects (age, 5 40 years; mean, 25 years), 2 patients with achromatopsia (ages 15 and 24 years), and 11 patients with mutations in the RPE65 gene (age, 5 30 years; mean, 17.4 years) were studied. Testing of achromatopsia patients was performed to validate the potential to reliably separate cone responses from rod responses. Informed consent was obtained from all subjects after explanation of the nature and possible consequences of the study. The study was approved by the University of Giessen Institutional Review Board for human subjects and was undertaken according to the tenets of the Declaration of Helsinki. Light Stimulus The ColorDome desktop Ganzfeld apparatus (Diagnosys LLC, Littleton, MA) was used as light-emitting source, which produces a diffuse widefield stimulus with a light-emitting diode. The photopically matched stimuli chosen for this study were nm (red light) and nm (blue light) (specifications of light-emitting diodes used provided by the distributor). Custom-made protocols were defined by free choice of stimulus color, intensity, stimulus duration, and interstimuli breaks. Pupillometer The pupillometer (Bino1; AMTech, Dossenheim, Germany) is a custommade advancement of the commercially available uniocular f2d apparatus. It enables separate stimulation of one eye or stimulation of both eyes at the same time, and simultaneous recording of both pupils. The pupillometer consists of two infrared video cameras to measure the pupil diameter. The subject carries the recording device and is positioned directly in front of the Ganzfeld stimulator (Fig. 1). Stimulus Protocols The original protocol for chromatic pupillometry has been published elsewhere. 47 Briefly, in our setting the measurements were performed under mesopic conditions, and pupil diameter recording was started 5 seconds before the first red light stimulus at 0 log cd/m 2 for 13 seconds, followed by a second red light stimulus at 1 log cd/m 2 for 13 seconds, followed by a third red light stimulus at 2 log cd/m 2 for 13 seconds. After 60 seconds of darkness, the stimulus cycle was repeated with blue light. The pupil diameter was recorded up to at least 60 seconds after the last light stimulus. The advanced protocol has also been published in detail elsewhere. 48 However, we modified it to minimize the discomfort experienced by our healthy probands, and to decrease difficulties in

3 IOVS, August 2012, Vol. 53, No. 9 Chromatic Pupillometry in RPE65 Deficiency 5643 measurement due to excessive blinking. We reduced the maximum stimulus intensity for red and blue stimuli from 2.6 log cd/m 2 to 2 log cd/m 2. Measurements of the rod-weighted answers and the iprgc answers were started after 10 minutes of dark adaptation. The measurement of the pupil diameter started 5 seconds before the first 2 log cd/m 2 blue stimulus was presented for 1 second, assessing rods, repeated three times with a 15-second dark period in between, and followed by three 2 log cd/m 2 blue stimuli for 1 second, assessing iprgcs, with 30 seconds darkness in between stimuli. The pupil diameter was recorded until at least 60 seconds after the last light stimulus. For the cone-weighted protocol, a blue background of 0.78 log cd/ m 2 was applied for 3 minutes before the measurement. The measurement started 5 seconds before the first 2 log cd/m 2 red stimulus for 1 second, and was repeated three times with a 15-second dark period in between, followed by three 2 log cd/m 2 blue stimuli for 1 second, with 30 seconds darkness between the stimuli. The pupil diameter was recorded up to at least 60 seconds after the last light stimulus. Four normal subjects (n ¼ 8 eyes) were examined twice with one week in between. Pupil Recording and Analysis The video signal (25 Hz) was relayed to a processing board that recorded the pupil diameter in real time into a text file. In all experiments, only one eye was stimulated, while the other eye was covered, but both pupil diameters were measured simultaneously. The results were exported and analyzed by custom-made computer software developed in our laboratory (Zahn S, Strohmayr E, Stieger K, Lorenz B, et al., manuscript in preparation). Raw data were preprocessed, which included removal of invalid measuring points and smoothing of the graph by using a polynomial fit (Savitzky-Golay filter, fourth order). Relative sustained and transient pupil response data (i.e., pupil baseline size minus pupil size after contraction divided by pupil baseline size, times one hundred) were extrapolated into box plots for better readability and analysis of significance. Statistical Analysis Statistical analysis was performed with Sigma plot 10 (Systat Software Inc., San Jose, CA). Groups with Gaussian distribution were analyzed with the Student s t-test, groups without Gaussian distribution with the Mann-Whitney rank sum test. Test retest variability was measured in 4 healthy subjects (n ¼ 8 eyes) by using two experimental protocols: (1) measurements were done at three different days within two weeks (9.00 AM); and (2) measurements were done at three different time points during the same day. Variability parameters included calculation of the mean value (with 95% confidence interval), the intraclass correlation coefficient (ICC), and the significance of test variability, using the two-tailed t-test or the Mann-Whitney rank sum test for the intraday experiment, and the ANOVA on ranks test for the intervisit experiment. Full-Field Stimulus Testing FST was performed as described previously for obtaining psychophysical estimates of light sensitivity. 27 Briefly, light stimuli were delivered with a ColorDome Ganzfeld stimulator. In this setup, stimulus luminance can be set within a range spanning 80 db for red, blue, or white light. For establishing normative data, a calibrated 3 log unit (30 db) neutral density filter was intercalated. Relative spectral content for the light sources was measured with a spectrometer (PR650; Photo Research Inc., Chatsworth, CA) for red and blue light, and luminance of the white stimulus and its spatial uniformity were measured with a directional photometer (IL1700; International Light, Peabody, MA). For each session, repeated measurements of sensitivity to a fullfield stimulus were obtained on dilated eyes in the dark-adapted FIGURE 1. Experimental setup. (A) The goggle (recording device) contains two infrared cameras for bilateral pupil recording at 25-Hz recording rate. Owing to the front of the goggle, which can be closed unilaterally, any given light source can be used to emit light of defined wavelength and intensity. (B) The subject is positioned in front of the Ganzfeld stimulator (Color Dome). The subject carries the recording apparatus (Bino I). (C) Output screen of the software program of the binocular chromatic pupillometer (Bino I), which allows real-time observation of the recorded pupil, the actual measured pupil diameters, and the line of sight of both eyes simultaneously. state (tropicamide eye drops, 45 minutes), with white, red, and blue flashes of 200-ms duration. A short pause was given between sensitivity determinations to avoid fatigue. Testing was performed at similar times of the day in all subjects. The starting luminance for thefirstsamplewasusuallychosentobeatleast10dbdimmerthan the subject s expected threshold, to avoid light-related rod desensitization. A total of 7 eyes were included in the healthy control group. Clinical Examination Further clinical parameters included in this study were BCVA using Snellen charts, and kinetic perimetry (Goldmann visual field). Planimetry calculations of the visual field were done as previously published. 53 RESULTS Custom-Made Pupillometry System and Test Retest Variability In our setting, a ColorDome Ganzfeld stimulator was connected to a binocular pupillograph, which measures the pupil diameter in both eyes simultaneously at 25 Hz. As the cameras are installed inside the goggles and the front for each eye can be covered separately (Fig. 1B), we can measure simultaneously the pupil responses of both eyes to unilateral stimulation. Measurement protocols were adapted from Kardon and colleagues 47 (original protocol, OP) and from Park and colleagues 48 (advanced protocol, AP). Figure 2 shows the distribution of accumulated data obtained from all eyes of the control group (gray), and also from one single control eye (black line), and from the left eye of RPE65 patient 11 (red line). The pupil responses were visualized as graphs (Figs. 2A, 2C, 2E) or box plots (Figs. 2B, 2D, 2F). For both protocols, transient (t) and sustained (s) pupil diameter values were measured (see Methods for definition), and those data defined as rod-, cone-, or iprgc- mediated pupil responses for both protocols were highlighted. Because box plot graphs allow for rapid visualization of significant alterations of the transient or sustained pupil response, only the corresponding box plots are shown in subsequent figures. Test retest variability was measured in 10 eyes from 5 healthy subjects in two experimental protocols to discern intervisit and intraday variability (Tables 1 and 2). ICC for all parameters was in the range of 0.34 to 0.78 for the intervisit experiment, and between 0.25 and 0.83 for the intraday experiment. Test values between examinations did not differ significantly.

4 5644 Lorenz et al. IOVS, August 2012, Vol. 53, No. 9 FIGURE 2. Experimental protocols for chromatic pupillometry. (A, B) Graph and box plots obtained with the OP. Transient pupil response (B) represents the maximum pupil constriction at 1 second after stimulus onset (indicated by the green triangle in [A]). Sustained pupil response (B) represents the maximum pupil constriction at the last second before stimulus change or offset (indicated by the yellow triangle in [A]). (C F) Graphs and box plots obtained with the AP for rod- and iprgc-weighted stimulation (C, D) or cone-weighted stimulation (E, F). Transient pupil response (D, F) represents the maximum pupil constriction at 1 second after stimulus offset (indicated by the green triangle in [C, D]). Sustained pupil response (D, F) represents the maximum pupil constriction at 6 seconds after stimulus offset (indicated by the yellow triangle in [C, E]). (A) Relative pupil constriction of the healthy control group (gray), one healthy subject, and the left pupil of RPE65 patient 11. (B) Box plots of transient and sustained pupil responses of the healthy control group data displayed in (A) (n ¼ 24 eyes), the healthy subject (as dark spot), and the left eye of RPE65 patient 11 (as red spot). Rod-weighted responses are defined as the transient values for 0 log cd/m 2 blue light stimulus. Cone-weighted responses are defined as the transient values for 2 log cd/m 2 red light stimulus. iprgc-weighted responses are defined as the sustained values for 2 log cd/m 2 blue light stimulus. (C) Relative pupil constriction of the healthy control group (in gray), one healthy subject, and the left pupil of RPE65 patient 11. (D) Box plots of transient and sustained pupil responses of the normal control group data displayed in (C), the healthy subject (as dark spot), and the left eye of RPE65 patient 11 (as red spot). Rod-weighted responses are defined as the transient values for 2 log cd/m 2 blue light stimulus. iprgc-weighted responses are defined as the sustained values for 2 log cd/m 2 blue light stimulus. (E) Relative pupil constriction of the healthy control group (in gray), one healthy subject, and the left pupil of RPE65 patient 11. (F) Box plots of transient and sustained pupil responses of the normal control group data displayed in (E), the healthy subject (as dark spot), and the left eye of RPE65 patient 11. Cone-weighted responses are defined as the transient values for 2 log cd/m 2 red light stimulus with 0.78 log cd/m 2 blue background. s, sustained; t, transient.

5 IOVS, August 2012, Vol. 53, No. 9 Chromatic Pupillometry in RPE65 Deficiency 5645 TABLE 1. Intervisit Variability Statistics Descriptive Statistics OP Rods AP Rods OP Cones AP Cones OP iprgc AP iprgc CA (%) CA (%) CA (%) CA (%) CA (%) CA (%) Mean (95% CI) Test No (39 50) 30 (29 32) 56 (53 59) 38 (35 41) 58 (53 62) 35 (30 41) Test No (40 49) 30 (27 33) 59 (56 62) 37 (35 39) 56 (52 61) 36 (32 41) Test No (39 49) 31 (28 34) 58 (55 61) 35 (33 27) 56 (53 60) 38 (34 42) Test retest analysis ICC (95% CI) 0.78 ( ) 0.64 ( ) 0.74 ( ) 0.34 ( ) 0.76 ( ) 0.45 ( ) ANOVA on ranks CA, contraction amplitude; CI, confidence interval. Differentiation of Cone-Weighted Pupil Response in Patients with Achromatopsia Transient pupil responses to the strongest red light in OP have been attributed to cone function, but this theory was challenged from the beginning, leading to significant modifications in AP 48 (see Methods section). To analyze both stimulus paradigms for cone response, we measured coneweighted pupil responses by using both protocols in two patients with complete achromatopsia associated with mutations in the CNGB3 gene (Figs. 3C, 3D). Clinical examinations confirmed complete color blindness (data not shown). Pupil responses were only slightly reduced in both patients when using OP (Fig. 3C). Interestingly, the pupil response from the right eye of patient 1 (A1) was even within the normal range of healthy controls. In contrast, pupil responses were severely reduced in AP, in accordance with absence of cone responses (Fig. 3D). As expected, pupil responses for the rod and iprgc stimulus paradigms were within the normal range (Figs. 3A, 3B, 3E, 3F). Chromatic Pupillometry in RPE65 Patients A total of 11 compound heterozygous patients with mutations in the RPE65 gene were included in this study (Table 3). The age was between 5 and 30 years. Visual acuity ranged from hand motion (HM) to 20/20 BCVA, and visual fields varied from 0.5% age-matched visual field for V/4e to normal visual field for III/4e (see Figs. 5G, 5H; Table 3), demonstrating the huge phenotypic variability in RPE65 deficiency that is in part due to the progressive nature of the disease. All patients had absent or severely reduced fundus autofluorescence and nonrecordable ERGs (data not shown). All patients underwent the OP (n ¼ 18 eyes) and six patients, the AP (n ¼ 10 eyes). Eyes were recorded separately. The significance of the results was analyzed for both protocols, but only the results from AP followed a Gaussian distribution, allowing to perform a Student s t-test (P < 0.001). The results obtained with OP were analyzed with a Mann-Whitney rank sum test (U-test, P < 0.013). Pupil responses to rod-weighted stimuli revealed significant differences between healthy control eyes and eyes from patients with RPE65 mutations (Figs. 4A, 4B). With OP, pupil responses were 40% to 50% in the control group and between 10% and 30% in the patient group. In contrast, with AP, pupil responses were between 30% and 35% in the control group and between 0% and 5% in the patient group. Pupil responses to cone-weighted stimuli revealed also significant differences between control and patient groups (Figs. 4C, 4D). With OP, pupil responses were 50% to 57% in the control group and between 30% and 55% in the patient group. With AP, pupil responses were between 30% and 35% in the control group and between 17% and 22% in the patient group. Pupil responses to iprgc-weighted stimuli revealed marginal, but significant differences between control and patient group (Figs. 4E, 4F). With OP, pupil responses were 50% to 60% for the control group and between 42% and 57% in the patient group. With AP, pupil responses were 30% and 45% for the control group and between 20% and 35% in the patient group. Rod-Weighted Data in Chromatic Pupillometry Correlate with the Clinical Phenotype Following these intriguing results, we evaluated the individual results of rod-weighted chromatic pupillometry against data on visual field and sensitivity differences in FST testing (Table 3, Fig. 5). FST is a dark-adapted psychophysical testing method in which the light sensitivity thresholds for chromatic (blue, red) and white stimuli are defined through presenting light stimuli with increasing intensity until the patient perceives them. 26 In healthy subjects under dark-adapted conditions, light sensitivity for blue light stimuli is much higher than for red stimuli (20 30 db), indicating a higher sensitivity of rods (detecting blue light) compared to cones (detecting red light). Consequently, a reduction of the difference below 20 db indicates a mixed response of rods and cones for the blue light stimulus, and differences below 3dB indicate pure cone-mediated vision. 3 Sensitivity loss to white light indicates a generally TABLE 2. Intraday Variability Statistics Descriptive Statistics OP Rods AP Rods OP Cones AP Cones OP iprgc AP iprgc CA (%) CA (%) CA (%) CA (%) CA (%) CA (%) Mean (95% CI) Test No (39 46) 31 (27 35) 57 (54 60) 35 (33 37) 58 (54 62) 38 (34 42) Test No (39 45) 27 (24 30) 57 (54 40) 35 (33 38) 56 (52 60) 39 (33 45) Test-retest analysis ICC (95% CI) 0.82 ( ) 0.45 ( ) 0.76 ( ) 0.25 ( 0.12 to 0.62) 0.83 ( ) 0.74 (0.44 1) P value (two-tailed t-test) P value (rank sum test)

6 5646 Lorenz et al. IOVS, August 2012, Vol. 53, No. 9 Cone-Weighted Data in Chromatic Pupillometry Do Not Correlate with the Clinical Phenotype We evaluated cone-weighted pupil responses against BCVA and white light sensitivity loss in FST. In cone-weighted OP, P2 (BCVA 20/20, no white sensitivity loss) exhibited pupil responses within the normal range, while the results of P5 (BCVA 20/40, no white sensitivity loss) remained slightly below normal (Fig. 5C). The slightly reduced cone-mediated pupil responses in P5 were confirmed with AP, in which a reduced but measurable pupil response was observed. The remaining patients, all with BCVA below 20/100 and severe white sensitivity loss of more than 40 db compared to normal (Fig. 5F), exhibited varying results in OP: some of them were within normal range, while others were severely reduced (Fig. 5C). Because we showed that a mixed activation of the pupil response is obtained with the cone-weighted OP in patients with achromatopsia (Fig. 3), the results from our patients with RPE65 mutations in cone-weighted OP may also be considered mixed responses. Pure cone answers, however, were observed with the cone-weighted AP, and all but two patients exhibited severely reduced pupil responses (Fig. 5D). However, the latter two patients (P4 and P8) had a very reduced BCVA of 20/400 and 20/1000, respectively, but revealed a stronger pupil reaction than the patient with the much better BCVA (P5, 20/40) (Table 3). The reason for this discrepancy remains unexplained. FIGURE 3. Chromatic pupillometry in achromatopsia. Cone-weighted pupil responses are present in OP, but absent in AP, indicating a mixed response in OP and an isolated cone response in AP. Rod- and iprgcweighted pupil responses remain unchanged. (A, C, E) Results of two patients with complete achromatopsia obtained with the OP and plotted against the box plots based on the data from the healthy control group. Rod-weighted responses are defined as the transient values for 0 log cd/m 2 blue light stimulus. Cone-weighted responses are defined as the transient values for 2 log cd/m 2 red light stimulus. iprgc-weighted responses are defined as the sustained values for 2 log cd/m 2 blue light stimulus. (B, D, F) Results of two patients with complete achromatopsia obtained with the AP and plotted on the box plots of the healthy control group. A1, patient 1; A2, patient 2; OD, right eye; OS, left eye. reduced light sensitivity for rods and cones. It is of note to mention that all patients had nonrecordable ERGs (data not shown). Two patients (P2 and P5) were analyzed separately because they presented with a very mild phenotype at age 9 years and 18 years, respectively, with a normal (P2) or 50% reduced (P5) visual field for the isopter III/4e (Fig. 5H, Table 3), and a higher sensitivity threshold for blue light than for red light FST (difference blue-red: >15 db for P2, >25 db for P5) (Fig. 5E). In addition, both did not exhibit sensitivity loss to white light in FST (Fig. 5F). These data clearly indicate the presence of rodmediated vision. The clinical phenotype correlated well with the near-normal pupil response of P2 and the slightly reduced response of P5 in the rod-weighted OP (Fig. 5A), and the moderately reduced pupil response of P5 in the rod-weighted AP (Fig. 5B). All other patients exhibited severely reduced or absent rod function in FST (Figs. 5E, 5F) and severely reduced visual fields (Figs. 5G, 5H; Table 3). In line, all those patients exhibited absent or severely reduced pupil responses to the rodweighted AP (Fig. 5B). In contrast, pupil responses in the rod-weighted OP were more heterogeneous, albeit generally reduced (Fig. 5A), indicating a mixed response from lightsensitive retinal cells. Correlation of Cone versus Rod-Mediated Pupil Response To further demonstrate the utility of chromatic pupillometry for dissecting function of rods and cones in healthy and diseased retinae, we plotted the cone-mediated pupil response against the rod-mediated pupil response for the healthy control groups, the achromats, and the RPE65 patients tested with OP and those tested with AP (Fig. 6). The healthy control group forms a cloud with both protocols, gathering around the 458 axis, which would indicate equally strong rod- and conemediated pupil responses (indicated by solid circles). Interestingly, the cloud for AP seems to have slightly more variability in both directions. However, when analyzing the data of the patient groups, it becomes clear that the AP protocol very clearly separates achromats as well as patients with RPE65 deficiency from the healthy control cloud (indicated by spotted circles), while the OP protocol does not for both patient groups (indicated by dotted circles). For example, both patients with achromatopsia can be found with AP close to the x-axis, indicating almost absent cone responses, while with OP, both patients can be found within the cloud of the healthy control group. Furthermore, while even the mild phenotype of RPE65 deficiency (P5) is clearly separated from the healthy control cloud with AP, some severe phenotypes of RPE65 deficiency (P9, P6) are located within the healthy control cloud with OP. Nevertheless, the mildest phenotype included in this study (P2) is located within the healthy control cloud with OP, which would be expected. Postillumination Pupil Response We measured the pupil diameter at thirty seconds after the last strong blue light stimulus offset with both protocols in order to characterize the PIPR (Fig. 7). Significant differences, compared to the healthy control group, were observed with both protocols. On average, with OP, the percentage of pupil constriction in the healthy control group was 3%, while the patient group exhibited a pupil constriction of 15% (medium

7 IOVS, August 2012, Vol. 53, No. 9 Chromatic Pupillometry in RPE65 Deficiency 5647 TABLE 3. Genotype and Phenotype of the 11 Patients with RPE65 Mutations Enrolled in This Study Patients Mutations Visual Acuity* Visual Field (%) Comments Age (y) Patient p.r91w/p.e102x OD 20/285 OD: V/4e OS 20/330 OS: V/4e 12.8 Patient p.p25l/p.p25l OD 20/20 OD: III/4e normal 9 OS 20/20 OS: III/4e normal Patient p.r91w/p.e102x OD 20/500 OD: V/4e 45.1; III/4e 0.7 OS: III/4e temporal island 2% 9 OS 20/400 OS: V/4e 43.3; III/4e 4.8 Patient c.11þ5g>a/p.y368h OD 20/400 OD: V/4e 52.6; III/4e OS 20/200 OS: V/4e 45.6; III/4e 33.9 Patient p.p25l/p.y79h OD 20/40 OD: V/4e normal; III/4e OS 20/40 OS: V/4e normal; III/4e 58.6 Patient p.r91w/p.y368h OD 20/100 OD: V/4e 8.4; III/4e OS 20/100 OS: V/4e 10.6; III/4e 2.3 Patient c.11þ5g>a/p.y368h OD 20/1000 OD: V/4e 16.2; III/4e 1.8 OD: V/4e temporal island 2.2% 20 OS 20/160 OS: V/4e 18.6; III/4e 1.9 OS: V/4e temporal island 8.3% Patient c.11þ5g>a/c.144inst OD 20/1000 OD: V/4e OS 20/160 OS: V/4e 5.3 Patient p.r91w/p.y368h OD 20/1000 OD: V/4e 13.5 OD: V/4e temporal island 11.6% 22 OS 20/125 OS: V/4e 8.7 OS: V/4e temporal island 5% Patient p.e102x/p.e102x OD: HM OD: V/4e OS: LP OS: V/4e 0.4 Patient c.11þ5g>a/c.725þ2t>a OD 20/200 OD: V/4e 5.6; III/4e 0.5 OD: V/4e temporal island 5.6% 30 OS 20/400 OS: V/4e 13.0; III/4e 0.2 LP, light perception; V/4e and III/4e, diopter targets. * BCVA. Visual field area as proportion of normal for age. Patients are siblings. Patients are siblings. value, P < 0.01) (Fig. 7A). Similarly, with AP, average pupil constriction was 5% in the healthy control group, while a residual constriction of 20% (medium value, P < 0.01) was observed in the patient group (Fig. 7B). These observations indicate an increased PIPR in patients with RPE65 deficiency. DISCUSSION In this study, an objective measurement technique was presented for the characterization of the light-sensitive retinal cell populations in patients with mutations in the RPE65 gene. We applied two protocols previously published by Kardon et al. 47 and refined by the same group, Park et al. 48 Isolation of cone responses was only realized with the AP. This was shown in patients with achromatopsia and proven absence of cone function, who erroneously exhibited pupil responses only when using the OP cone-weighted protocol. In general, all patients with RPE65 mutations exhibited significantly reduced or absent rod-weighted pupil responses, and reduced iprgc function. The reduced rod function correlated well with the clinical phenotype, measured by visual field planimetry and FST. Cone-mediated pupil responses were also reduced, but independent from BCVA, as patients with fairly good residual vision responded poorly and patients with very low BCVA responded quite well. Test retest variability in the healthy control group was within an acceptable range without significant differences between the test values, regardless of whether tests were performed the same day (intraday variability) or at several visits (intervisit variability) (Tables 1 and 2). This observation goes well in line with variability reported by other groups. 54 In any case, when analyzing the treatment benefit of experimental therapies for RPE65 deficiency in individual patients, several pretest examinations will be necessary to define the individual baseline variability. FIGURE 4. Chromatic pupillometry in the patient group with RPE65 deficiency. Pupil responses for all three conditions are significantly reduced with both protocols. (A, C, E) Box plot data obtained with the OP. Rod-weighted responses are defined as the transient values for 0 log cd/m 2 blue light stimulus. Cone-weighted responses are defined as the transient values for 2 log cd/m 2 red light stimulus. iprgc-weighted responses are defined as the sustained values for 2 log cd/m 2 blue light stimulus. (B, D, F) Box plot data obtained with the AP. (A, B) Box plots obtained for rod-weighted stimuli. (C, D) Box plots obtained for coneweighted stimuli. (E, F) Box plots obtained for iprgc-weighted stimuli. *U-test, P < 0.013; **t-test, P <

8 5648 Lorenz et al. IOVS, August 2012, Vol. 53, No. 9 FIGURE 5. Chromatic pupillometry, FST, and Goldmann perimetry in all patients with RPE65 deficiency displayed separately. (A, B) Rod-weighted pupil response data obtained with (A) OPand(B) AP for all patients separately. (C, D) Cone-weighted pupil response data obtained with (C) OPand(D) AP.(E, F) FST data of all patients separately. The difference in sensitivity threshold for blue versus red light under dark-adapted conditions is displayed in (E). Data for the healthy control group (n ¼ 7 eyes) are shown for comparison. Differences above 20 db indicate rod-mediated responses to the blue stimulus. Differences between 3 and 20 db are considered to be mixed rod cone responses, and differences below 3 db indicate pure cone-mediated perception of the blue stimulus. 3 (F) Sensitivity loss to a white stimulus obtained for all patients separately were compared with a healthy control group (n¼7eyes).(g, H) Goldmann visual field data for isopters (G) V/4e and (H) III/4e. Visual field areas of all patients separately are displayed as percentage of normal for age. It is known that rod function is severely compromised at birth or within the first years of life in patients with mutations in the RPE65 gene, which is due to the complete dependence of rods on 11-cis retinal supply from the RPE-based visual cycle. 9,10,55 This goes in line with the typical phenotype of severely reduced light sensitivity and visual field at early ages. 3 Reduced or absent pupil responses to rod-weighted stimuli in chromatic pupillometry therefore correctly correlated with this observation (Figs. 4A, 4B, 5, 6). Similarly, patients with RP and reduced visual field due to loss of rod-mediated vision also display a reduced rod-weighted answer with the OP. 56 Furthermore, RP patients with the strongest reduction in visual field and absence of ERG recordings exhibit most reduced rod-weighted pupil responses. Interestingly, it has been observed that patients who undergo AAV-mediated gene therapy for correction of RPE65 deficiency display an increased light sensitivity threshold, as measured by white or blue light FST. 22,23 This supposedly rodmediated rescue effect should also be observable when using chromatic pupillometry and could therefore be quantified by using the objective measurement technique presented in this article. It is of note, however, that chromatic pupillometry is a summation response of the entire retinal area to a light stimulus. Therefore, analysis of local treatment effects in AAV-

9 IOVS, August 2012, Vol. 53, No. 9 Chromatic Pupillometry in RPE65 Deficiency 5649 FIGURE 6. Correlation of cone- versus rod-mediated pupil responses with OP and AP in healthy control subjects, achromats, and RPE65 patients. Solid black circles indicate the cloud of healthy control subjects examined with OP and those examined with AP. Dotted black circles indicate the clouds for achromats and RPE65 patients examined with AP. Spotted black circles indicate the clouds for achromats and RPE65 patients examined with OP. ACH, achromats. mediated gene therapy trials will continue to be measured by using spatial resolution analysis, such as dark-adapted twocolor threshold perimetry. Nonetheless, the objective chromatic pupillometry can add important functional data when it comes to rescue effects that affect the PLR. In addition, current clinical trials using the oral substitution of retinoid compounds are likely to have an effect on the entire retina, thus making local functional differences less likely to appear (Koenekoop R, et al. IOVS 2011;52:ARVO E-Abstract 3323). The involvement of cones in the pathology of RPE65 deficiency is less well understood. Early S-cone function loss has been described in humans 11 (Lorenz B, Stark R, Wegscheider E, et al., manuscript in preparation), and early loss of S-cone gene expression was observed in RPE65 knockout mice. 57 M/L cones seem to be less affected and mediate latestage vision in human patients, as demonstrated by FST and chromatic sensitivity measurements 3,11 (Lorenz B, Stark R, Wegscheider E, et al., manuscript in preparation). Recent FIGURE 7. PIPR in patients with RPE65 deficiency. The PIPR is significantly increased in the patient group. (A) Box plot data obtained with the OP at 30 seconds after the last stimulus offset. (B) Box plot data obtained with the AP at 30 seconds after the last stimulus offset. *U-test, P < findings demonstrating RPE65 expression in L/M cones in mice would explain differing survival rates among cone subtypes, as loss of chromophore supply would only affect rods and S- cones. 6,7 Interestingly, L/M cone opsin delocalization has been observed in RPE65 knockout mice at early time points (P25), 58 while cone degeneration in RPE65 R91W knockin mice is much slower. 59 Using chromatic pupillometry, cone function was reduced in most patients (Figs. 4C, 4D), but did not correlate well with VA or white light sensitivity threshold in FST (Figs. 5C F, Table 3). The reason for this remains elusive. Cone-weighted stimulation using OP was shown to be a mixed response in achromats (Fig. 3). However, rods can only play a minor role, as most of the patients exhibited lack of rod mediation (Figs. 5E, 5F). A contribution of iprgcs to the mixed response is possible, albeit not very likely, given the brightness of the red light stimulus at 640 nm and the chromatic sensitivity of iprgcs. 45 The inconclusive data concerning the cone answer, using the AP and VA data, are also difficult to explain, as examinations in achromats demonstrated a fairly good isolation of cone responses (Figs. 3, 6). A possible explanation would be that the PLR integrates the entire retinal input of cones, while VA is mainly determined by the cone population in and near the fovea. Similarly, with subjective FST data, it may be that a person still cannot integrate the entire retinal cone input to the extent the pupil system does. Or, the protocol readouts are very close to the threshold values, and borderline effects artificially magnify small differences. The participation of the iprgcs to the PLR is currently under debate. Studies in mice suggest an interaction of the outer retinal visual cycle and the melanopsin-driven visual cycle in ganglion cells, because mice lacking RPE65 or lecithin retinol acyltransferase (LRAT) possess PLRs with reduced light sensitivity. 41,60,61 However, the same studies also support an independence of the melanopsin system from RPE65 activity, leaving this question largely unsolved. Doyle and colleagues 61

10 5650 Lorenz et al. IOVS, August 2012, Vol. 53, No. 9 have argued for a mechanism by which rods may influence the function of iprgcs. In humans, patients without any light perception for bright red and low blue light stimuli display a pupil constriction to bright blue light stimuli, 62 which is considered to be iprgc mediated. In addition, a patient with Leber hereditary optic neuropathy and severely reduced photoreceptor-mediated vision possesses an almost normal PLR to bright blue light stimuli, which is explained by a sparing of iprgc from ganglion cell loss in this disease. 63 In any case, complete isolation of iprgc contribution to the PLR in a clinical setup will likely remain a difficult task, as the absorption maximum is close to that of rods and S-cones, and S-cones certainly contribute to the PLR following bright blue light stimulation. Consequently, the possible reduction or absence of S-cone function in patients with RPE65 deficiency may be the cause for the reduced iprgc response in chromatic pupillometry (Figs. 4E, 4F). On the other hand, achromats with complete absence of cone function did not exhibit a reduced PLR to the bright blue light stimulus (Fig. 3C). A different way to study the iprgc contribution is to evaluate the PIPR after bright blue light stimulus offset. The pupil redilates differently under these conditions compared to bright red light stimulus offset. 64 The slow kinetics is associated with continuous iprgc firing, which eventually discontinues, leading to redilation. 46 It has been shown in glaucoma patients that the PIPR decreases with advancement of the disease, indicating a loss of iprgcs in this pathology. 49,50 Similar results have been observed in patients with diabetic retinopathy. 51 In our patient cohort with mutations in the RPE65 gene, the redilatation kinetics was slower than that of the healthy control group (Fig. 7), which means that the PIPR was increased. This continued firing of the iprgcs may be related to the absence of rod signaling. 65 With OP, increased redilatation times are also observed in patients with RP 56 or with unclassified outer retinal disease, 66 which would support this hypothesis. On the other hand, S-cone interference has also been associated with iprgc firing. 46 In particular, Allen and colleagues 67 have recently demonstrated S-cone mediated off inhibition of firing in the pretectal olivary nucleus in mice, indicating an important contribution of this photoreceptor type to nonimage-forming vision. Consequently, a lack of S-cone function may also cause an increased PIPR. Current examination techniques that generate useful information in patients with lowest levels of residual vision, such as patients with RPE65 mutations, include exclusively psychophysical testing methods such as FST or funduscontrolled, two-color threshold perimetry. In all these techniques, cooperation of patients is crucial, and variations in testing results due to less cooperative habits cannot be excluded. In addition, examination of the clinical benefit following experimental treatment methods remains at least in part subjective owing to active participation of the patient. Pupillometry, on the other hand, is absolutely objective, as the PLR depends on active neuronal projections from the retina to the brainstem and back to the eye, and thus cannot be modified by the patient. Of course, the emotional status of the patient at the time of the examination, as well as fatigue, may have an impact on the performance, but this could be largely excluded by examining the patient beforehand and taking pauses between examinations. Pupillometry using white or green light at increasing intensities displays a summation of the responses of all light-sensitive cell types within the retina. Chromatic pupillometry, on the other hand, has been developed to discern this summation response into its contributing elements. Of course, complete separation of the rods, cones, or iprgcs is difficult to achieve and cannot be expected even with the advanced protocol. Nonetheless, at least partial information about the different light-sensitive cell populations can be advantageous when estimating disease progression or therapeutic outcome. In summary, we presented the results of a new objective testing method for the functional analysis of rods, cones, and iprgcs in patients with RPE65 mutations, which should be added to the panel of examination tools for the characterization of the natural history of retinal degenerative disorders and for quantifying the clinical benefit in patients following experimental treatment protocols. Acknowledgments The authors thank Olga Meier for technical support during patient examination. Our thanks also go to all patients and their families participating in the study. References 1. Gu SM, Thompson DA, Srikumari CR, et al. Mutations in RPE65 cause autosomal recessive childhood-onset severe retinal dystrophy. Nat Genet. 1997;17: Lorenz B, Gyurus P, Preising M, et al. Early-onset severe rodcone dystrophy in young children with RPE65 mutations. Invest Ophthalmol Vis Sci. 2000;41: Jacobson SG, Aleman TS, Cideciyan AV, et al. Defining the residual vision in Leber congenital amaurosis caused by RPE65 mutations. Invest Ophthalmol Vis Sci. 2009;50: Lorenz B, Wabbels B, Wegscheider E, Hamel CP, Drexler W, Preising MN. Lack of fundus autofluorescence to 488 nanometers from childhood on in patients with early-onset severe retinal dystrophy associated with mutations in RPE65. Ophthalmology. 2004;111: Lorenz B, Poliakov E, Schambeck M, Friedburg C, Preising MN, Redmond TM. A comprehensive clinical and biochemical functional study of a novel RPE65 hypomorphic mutation. Invest Ophthalmol Vis Sci. 2008;49: Tang PH, Buhusi MC, Ma JX, Crouch RK. RPE65 is present in human green/red cones and promotes photopigment regeneration in an in vitro cone cell model. J Neurosci. 2011;31: Tang PH, Wheless L, Crouch RK. Regeneration of photopigment is enhanced in mouse cone photoreceptors expressing RPE65 protein. J Neurosci. 2011;31: Znoiko SL, Crouch RK, Moiseyev G, Ma JX. Identification of the RPE65 protein in mammalian cone photoreceptors. Invest Ophthalmol Vis Sci. 2002;43: Jin M, Li S, Moghrabi WN, Sun H, Travis GH. Rpe65 is the retinoid isomerase in bovine retinal pigment epithelium. Cell. 2005;122: Moiseyev G, Chen Y, Takahashi Y, Wu BX, Ma JX. RPE65 is the isomerohydrolase in the retinoid visual cycle. Proc Natl Acad Sci U S A. 2005;102: Jacobson SG, Aleman TS, Cideciyan AV, et al. Human cone photoreceptor dependence on RPE65 isomerase. Proc Natl Acad Sci U S A. 2007;104: Acland GM, Aguirre GD, Ray J, et al. Gene therapy restores vision in a canine model of childhood blindness. Nat Genet. 2001;28: Le Meur G, Stieger K, Smith AJ, et al. Restoration of vision in RPE65-deficient Briard dogs using an AAV serotype 4 vector that specifically targets the retinal pigmented epithelium. Gene Ther. 2007;14: