Comparison of acute non-visual bright light responses in patients with optic nerve disease, glaucoma and healthy controls

This study examined the effect of optic nerve disease, hence retinal ganglion cell loss, on non-visual functions related to melanopsin signalling. Test subjects were patients with bilateral visual loss and optic atrophy from either hereditary optic neuropathy (n = 11) or glaucoma (n = 11). We measured melatonin suppression, subjective sleepiness and cognitive functions in response to bright light exposure in the evening. We also quantified the post-illumination pupil response to a blue light stimulus. All results were compared to age-matched controls (n = 22). Both groups of patients showed similar melatonin suppression when compared to their controls. Greater melatonin suppression was intra-individually correlated to larger post-illumination pupil response in patients and controls. Only the glaucoma patients demonstrated a relative attenuation of their pupil response. In addition, they were sleepier with slower reaction times during nocturnal light exposure. In conclusion, glaucomatous, but not hereditary, optic neuropathy is associated with reduced acute light effects. At mild to moderate stages of disease, this is detected only in the pupil function and not in responses conveyed via the retinohypothalamic tract such as melatonin suppression.


S3
including electroretinography and neuroimaging were negative for any other cause of optic nerve damage. Inclusion for the patient group with a diagnosis of chronic open angle glaucoma (GL) was previous documentation of elevated intraocular pressure, bilateral visual field defects typical of glaucomatous visual loss without evidence of progression in the preceding year, and optic disc cupping and no other cause of optic nerve disease.
Glaucoma patients previously treated with trabeculectomy were excluded due to potential effect on iris structure and pupillary movement. Patients using topical agents with known effects on pupillary function, e.g. pilocarpine, brimonidine were also excluded. Diabetes and other neurologic deficit, for example hearing loss were exclusionary factors. A total of 1100 medical charts were screened for inclusion in the study. Due to the rigorous exclusionary criteria, only 50 patients were identified as potential study participants and were invited to fill out questionnaires and to undergo an ophthalmological examination. Twenty five of those patients agreed to participate in the study. Two patients did not complete the study due to acute sickness and one patient was excluded retrospectively due to current use of pain pills and alcohol abuse which were not stated at the time of interview. A total of 11 patients with hereditary optic nerve disease and 11 glaucoma patients were included in the final data analysis.
The 11 HON patients were four women and seven men aged 21 to 64 years (39.4 ± 15.2 years; mean ± SD; Table 1). None was taking a centrally acting medication. None of the HON patients was an extreme morning type, and PSQI scores ranged from 1 to 7 (4.4 ± 2.0), with three patients having PSQI scores greater than 5. The BDI was on average 1.7 ± 1.8 and ranged from 0 to 5. In 7 patients, results of gene testing were available from chart review. Three patients had a mutation of mitochondrial DNA associated with Leber hereditary optic neuropathy. In The GL patient group consisted of eight women and 3 men whose age ranged from 40 to 63 years (54.1 ± 7.1 years; mean ± SD; Table 1). Two patients were extreme morning types (HO scores>70); the PSQI scores ranged from 1 to 11 (mean ± SD: 5.5 ± 3.8) with four scores > 5, indicating some sleep related problems in these patients. The BDI was on average 2.1 ± 1.9 and range from 0 to 7. Visual acuity (VA) ranged from 0.05 and 1.0 (0.7 ± 0.2). Mean deviation (MD) ranged from 1.7 db to 24.2 db (11.4 ± 6.2) and mean peripapillary RNFL in the OCT was 59.8 ± 16.5 μm (range: 35 to 95 μm).

Age-matched controls
For the control group, healthy non-smoking volunteers were recruited via flyers in the region of Lausanne (Switzerland). Controls were matched to patients' age (± 3 years). All control participants were without psychiatric, medical or ocular disorders and not taking any prescription or non-prescription medications on a regular basis.
For control subjects, the inclusionary criteria from questionnaires included a PSQI to be lower or equal 5 (to exclude any sleep disorders), an HO score between 30 and 70 (to exclude extreme chronotypes) and a BDI less than 10 (to exclude for depression). All age-matched control participants had to have a normal ophthalmologic examination with no evidence of previous or current ocular disease other than refractive error. All controls had visual acuity of 1.0 or better (1.1 ± 0.1) and identified all 13 Ishihara color plates independently with each eye. The visual field of each control was judged to be normal and the MD for all controls ranged from -2.5 to 0.7 db (mean ± SD: -0.7 ± 0.8 db). Similarly, the OCT of controls was read as normal and the peripapillary RNFL measured 100.5 ± 11.3 μm, mean ± SD (range 68 to 125 μm). The HON control group ranged from 19 to 59 years and was composed of eight women, 3 men (age: 36.2 ± 13.2 years). The GL control group ranged from 42 to 63 years, with seven women and four men (54.4 ± 7.2 years). The demographic and ophthalmologic features of the patients and controls are presented in Table 1.

Salivary melatonin
Salivary samples for melatonin assays were obtained every hour and then immediately stored at 4° C. After study completion, the samples were centrifuged and frozen at -20° C before sending them to an external laboratory for radio-immunoassays (RIA; Dr. B. Middleton; University of Surrey; Guildford; UK). The inter-assay S6 coefficients of variance were 12.4% (low) and 8.5% (high). The intra-assay coefficients of variance were 6.9 % (low) and 2.4% (high) with a detection limit of 0.6 pg/ml.  There was no statistical difference between left and right eye pupil size (p>0.27 patients and p>0.1 controls), therefore pupil data from both eyes were averaged in all analyses. This was done to account for any potential differences that might occur from a difference in baseline pupil size, e.g. anisocoria. For two GL patients, eye movement artifacts precluded using data from one eye. A total of 6.4 %

Cognitive Performance
Two auditory-based cognitive performance tests were administered. Every hour, participants had to complete the 5-minute version of the Psychomotor Vigilance Task (PVT) 5 . In this task, the participant heard single tones and had to press the space key on the laptop as quickly as possible. A maximum of 50 tones were presented in random intervals. For the analysis, median reaction time (RT) and the 10% fastest and 10% slowest RT per trial were analyzed. Lapses, defined as RT > 500 ms were calculated separately, and RTs < 150 ms (anticipation) were not included in the analysis. The second performance test, the auditory n-back 6 was completed every two hours (five sessions). In this task, participants had to respond to spoken letters by pressing keys for correct or incorrect answers. In the 0-back test, the correct answer was when the participant heard the letter 'K' and pressed 'yes'; in the 2-back test the participant had to press 'yes' when the current letter which was played to the participant was identical with the penultimate one, otherwise the S9 participant had to press 'no'. In the 3-back test the participant had to press 'yes' if the current letter which was played to the participant was the same as the third last one, otherwise the participant had to press 'no'. The order of the letters was different for each n-back test and each test session; each of the five test sessions contained five 0-, 2-and 3-backs trials in a randomized order and in each trial a total of 30 letters were presented. The entire test lasted approximately 8 minutes. During the daytime screening visit and before the first test session in the evening, the participant was instructed and was trained with a demo-version, where feedback was given. During the test, the participants received no feedback on their performance. The results were analyzed by calculating accuracy as hits minus false alarms for each n-back version separately.

Statistics
Statistical analyses were performed by using the software packages SAS (SAS Institute Inc., Cary, NC, USA; v9.3 and Statistica v9). For single comparisons we applied two-tailed t-tests. For VAS, PVT and n-back tests, three GL patients were excluded from the analysis since they had reported use of sleep pills (two patients) and antihistamines (one patient) on a non-regular basis. Urinary toxicological screen for these three patients was however negative. Salivary melatonin, VAS, PVT and Nback data were analyzed with a mixed linear regression model (proc mixed) with the fixed factors='group' (patients vs. controls; separate for HON and GL patients); and the repeated factor 'time' (=time bins since study start; i.e. 10 hours for absolute and 3 hours on relative values since the beginning of LE), if not otherwise stated in the text. For the lapses in the PVT a non-parametric test (Mann-Whitney U Test) was used. The age was included as covariate in the analysis of cognitive performance tests (PVT and n-back) and subjective sleepiness (KSS and VAS). The analyses S10 were performed on log-or square root transformed data if the data was not normally distributed. VAS comparisons between groups were analyzed with relative data (differences to pre-light exposure). For KSS analyses the absolute data were ztransformed and plotted as difference relative to pre-light exposure. All p-values were adjusted for multiple comparisons with the Tukey-Kramer test and the degrees of freedom were adjusted (after Kenward-Rogers). The effect sizes (Cohen's d) were indicated for the melatonin and subjective sleepiness and PVT results in the text and were plotted for the pupil results in supplemental material ( Figure S2; d=2 small effect, d=0.5 medium effect and d=0.8 large effect). To examine the relationship between the PSPS and relative melatonin suppression a Spearman rank correlation analysis was performed.