Abstract
Neurodegenerative processes in the brain are reflected by structural retinal changes. As a possible biomarker of cognitive state in prodromal α-synucleinopathies, we compared melanopsin-mediated post-illumination pupil response (PIPR) with cognition (CERAD-plus) in 69 patients with isolated REM-sleep behavior disorder. PIPR was significantly correlated with cognitive domains, especially executive functioning (r = 0.417, p < 0.001), which was more pronounced in patients with lower dopamine-transporter density, suggesting advanced neurodegenerative state (n = 26; r = 0.575, p = 0.002). Patients with mild neurocognitive disorder (n = 10) had significantly reduced PIPR (smaller melanopsin-mediated response) compared to those without (p = 0.001). Thus, PIPR may be a functional—possibly monitoring—marker for impaired cognitive state in (prodromal) α-synucleinopathies.
Keywords
INTRODUCTION
The retina is considered part of the brain and has recently attracted attention as a possible indicator for neurodegenerative processes [1]. Structural retinal changes were reported for Alzheimer’s disease, Parkinson’s disease (PD), and Lewy body dementia, as well as for preclinical cognitive decline [2–7]. In patients with isolated REM-sleep behavior disorder (iRBD), a prodromal state of α-synucleinopathies, the retinal ganglion cell layer (GCL) is impaired and further deteriorates in the course of the disease [6]. GCL thinning was shown to be predictive for cognitive impairment in α-synucleinopathies [4]. A subtype of retinal ganglion cells, melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs), show a loss of synaptic contacts in PD [7]. Functionality of melanopsin-expressing ipRGCs can be analyzed using chromatic pupillometry [8, 9]. The aim of the present study was to investigate the relationship between functionality of ipRGCs and cognitive impairment in patients with iRBD, using chromatic pupillometry.
METHODS
Participants
Of 187 consecutive patients with video-polysomnographically confirmed iRBD, 118 patients were excluded from evaluation because of eye diseases (n = 38), diabetes (n = 12), sleep/wake promoting medication (n = 63) and technical reasons (n = 5). The included 69 patients performed chromatic pupillometry and cognitive testing between December 2015 and January 2021 in the Clinic for Sleep- & Chronomedicine in the St. Hedwig Hospital, Berlin, Germany. Of these patients, 51 underwent dopamine-transporter single-photon emission computed tomography (DaT-SPECT), 9 took antidepressants (no tricyclic antidepressants) and 20 took beta blockers.
All patients provided written informed consent for their anonymous clinical data to be used for analysis and publication. The ethics committee of Charité – Universitätsmedizin Berlin approved publication of post-hoc data analysis.
Pupil light reflex
Dark-adapted (10 min; 0 lux) pupil light reflex was recorded with a portable handheld device (NeuroLight, IDMED, Marseille, France) using the same protocol as previously described [10]. Minimum pupil size (MPS), mainly mediated by rods and cones [8], was defined as difference between baseline pupil size and maximal constriction after a 1-s blue light pulse (56 cd/m2) of monochromatic light with a peak wavelength at 465 nm. Post-illumination pupil response (PIPR), mainly mediated by melanopsin-expressing ipRGCs [8], was defined as difference between baseline pupil size and pupil size at 6 s after the light pulse. Compared to other PIPR metrics, the PIPR at 6 s after a 1-s blue light pulse has previously been reported to result in the largest PIPR amplitude with the lowest intra- and interindividual variability. And its spectral sensitivity could be described by the nomogram of the photopigment melanopsin, thereby making it a useful measure of melanopsin contribution to the pupil light reflex [11, 12].
Each pupil measurement was examined visually, and blinking artifacts were corrected using Excel’s linear trend smoothing option. Data was only included if there were no artifacts at the exact time of MPS or PIPR. To reduce circadian influences, cognitive testing and pupillometry were performed exclusively between 08 : 00 h and 15 : 00 h, never more than 30 minutes apart [13].
Cognition
Cognition was assessed using the CERAD-Plus test battery by trained staff (S.S. – psychologist). Two composite scores were formed: CERAD-composite score [14] and a score for executive functioning, which included subtests TMT-B, semantic fluency and phonemic fluency (equally weighted). Both composite scores were adjusted for age, gender and education [14]. Mild neurocognitive disorder (mNCD) was diagnosed using DSM-5 criteria by an experienced psychiatrist (D.K.).
DaT-SPECT
To identify elevated risk of developing α-synucleinopathies, striatal to occipital cortex uptake ratios of 185 MBq 123I-Fp-CIT (DaTSCANTM, GE Healthcare), administered 4 h prior to SPECT imaging, were calculated for both hemispheres, using BRASS-analysis (Hermes Medical Solutions, Sweden). This analysis references a normal image template, expressing deviations as an age-matched z-score. Lower DaT-density was defined if the uptake had a z-score >2.0 standard deviations below the reference, in the right or left posterior putamen [15]. DaT-SPECT was performed a median 2 days after pupillometry.
Statistics
All calculations were done with IBM SPSS Statistics for Windows, version 25.0. (IBM Corp., Armonk, NY, USA). Pearson correlations were performed for parameters which showed normal distribution, Spearman correlations for parameters without normal distribution. Linear mixed models were performed on data with normal distribution to assess group differences between patients with iRBD only and patients with iRBD + mNCD. Covariates age, sex, and time of recording were used for all analyses.
RESULTS
Participants
69 patients were included in the analysis (age: 67±9 years, mean±SD; range: 47–82 years; 14 female). In three patients MPS could not be determined due to blinking artifacts. Ten patients were diagnosed with mNCD; 26 patients had lower dopamine-transporter density (>2.0 SD below age norm).
Group comparisons
Compared to iRBD only, linear mixed models showed in iRBD + mNCD significantly reduced PIPR (i.e., faster redilatation; F1,69 = 13.014, p = 0.001) and MPS (i.e., less constriction; F1,66 = 7.075, p = 0.010; Fig. 1A). Covariates age, sex, and time of recording were not significant (p > 0.05).
Entire cohort of patients. A) Tracings of chromatic pupillometry. Dark blue line () indicates mean pupil size of patients with iRBD only; light blue line (
) indicates mean pupil size of patients with iRBD + mNCD, with 95% confidence interval error bars. Dotted lines show time points of MPS and PIPR. *p < 0.05; **p < 0.01. B–E) Correlations in entire cohort of patients. Correlations of PIPR with CERAD-composite score (B) and executive functioning (C). Correlations of MPS with CERAD-composite score (D) and executive functioning (E). n = 66 for MPS, due to blinking artifacts in 3 patients. Patients with iRBD only are shown in dark blue (
) and iRBD + mNCD in light blue (
). iRBD, isolated REM-sleep behavior disorder; mNCD, mild neurocognitive disorder; MPS, minimum pupil size; PIPR, post-illumination pupil response.
Correlations of pupillometry with cognition
PIPR was significantly and positively associated with both composite scores: global cognition (r = 0.292, p = 0.015; Fig. 1B) and executive functioning (r = 0.417, p < 0.001; Fig. 1C). Both composite scores did not correlate significantly with MPS (p > 0.05; Fig. 1D, E). Cognition subtests were significantly and positively associated with PIPR for word list learning, constructional praxis recall and TMT-B (p < 0.05), and showed trends in semantic fluency, and wordlist recall (p < 0.10). MPS was significantly and positively associated with TMT-A and TMT-B (p < 0.05).
In patients with lower DaT-density, PIPR was significantly and positively associated with both composite scores: global cognition (r = 0.544, p = 0.004; Fig. 2A) and executive functioning (r = 0.575, p = 0.002; Fig. 2B). In patients with normal DaT-density, PIPR showed no significant correlations: global cognition (r = 0.126, p = 0.548; Fig. 2C) and executive functioning (r = 0.225, p = 0.279; Fig. 2D). Both composite scores in both subgroups did not correlate significantly with MPS (p > 0.05).
Subgroup of patients with DaT-SPECT. Correlations of PIPR with CERAD-composite score and executive functioning. A, B) Patients with lower DaT-density. C, D) Patients with normal DaT-density. Patients with iRBD only are shown in dark blue () and iRBD + mNCD in light blue (). iRBD, isolated REM-sleep behavior disorder; mNCD, mild neurocognitive disorder; PIPR, post-illumination pupil response.
After exclusion of those 20 patients with beta blockers, analysis of the remaining 49 patients showed similar correlations between PIPR and both composite scores as in the total group (global cognition: r = 0.320, p = 0.025; executive functioning: r = 0.346, p = 0.015). Analyzing the 20 patients with beta blockers separately, the correlation of PIPR and executive functioning showed a trend (r = 0.438, p = 0.054), whereas PIPR and global cognition were not associated (r = –0.063, p = 0.793).
Covariates age, sex, and time of recording were not significant (p > 0.05).
DISCUSSION
The presented data suggest a substantial association between melanopsin-mediated PIPR and cognitive performance in patients in the prodromal stage of α-synucleinopathies. The association was more pronounced in those patients who seemed closer to the point of conversion, as indicated by lower DaT-density. Moreover, in iRBD patients who also had mNCD, PIPR was significantly more impaired (i.e., less light sensitive) than in iRBD patients without mNCD. Interestingly, the MPS, which is predominantly controlled by rods and cones, was not associated with cognitive performance. Thus, not impairment of the retina per se, but mainly deterioration of melanopsin-expressing ipRGCs might explain the link between pathological retinal changes and impaired cognition.
The CERAD Test battery was originally designed for patients with Alzheimer’s disease. To improve diagnostic validity in subcortical disorders, executive functioning tests (TMT-B and phonetic fluency) were included (CERAD-plus) [16]. In PD up to 39% of patients were diagnosed with mNCD, based solely on impairment of executive functioning [17]. Correspondingly, the association in our patients between PIPR and the composite score for executive functioning was stronger than the correlation with the more general CERAD-composite score. The associations remained after patients with beta blockers were excluded, indicating a significant correlation that persisted even after removal of one-third of the subjects.
Cognitive impairment in prodromal α-synucleinopathies is known to become pronounced within the last 6 years prior to conversion [18]. The most reliable marker of disease progression is dopamine-transporter density. Therefore, we took a closer look at iRBD patients with lower DaT-density since they may be in a more advanced prodromal state (i.e., closer to phenoconversion). The significant association we found between cognitive impairment and PIPR was especially strong in this group of patients. This finding may prove particularly valuable in the context of future disease modifying clinical trials. PIPR might serve as a functional biomarker to indicate individual treatment responses in advanced stages of prodromal α-synucleinopathies.
Our results are consistent with findings of structural impairment of retinal GCL in Lewy Body dementia and RBD [6]. Disruption of the GCL has been linked to cognitive impairment in PD [5] and predicted the risk of cognitive decline in PD [4]. In patients with multiple sclerosis a thinning of the GCL was shown to be associated with functional impairment of melanopsin-expressing ipRGCs, assessed with PIPR [19]. Accordingly, our findings suggest that this association between a thinning of the GCL and functional impairment of ipRGCs also exists in α-synucleinopathies.
The melanopsin-mediated PIPR reflects the functionality of melanopsin-expressing ipRGCs [8, 12]. We used the PIPR at 6 s after a 1-s light pulse in dark adapted patients [10, 20]. This method was recommended by Park et al. (2011) [11] and validated for α-synucleinopathies [21, 22]. Other methods of pupillometry exist which measure the steady-state pupil constriction during continuous light exposure [23, 24]. Both methods measure the melanopsin contribution to the pupil light response. For a review on the different methods see Kelbsch et al. (2019) [9].
Presented findings indicate that the dysfunctionality of melanopsin is linked to cognitive impairment in α-synucleinopathies, which could be explained via two neuronal pathways. First, it is known that ipRGCs have direct projections to brain areas responsible for higher cognitive functioning [25]. Secondly, ipRGCs provide information for the entrainment of the circadian system by projecting to the suprachiasmatic nucleus [26]. A disruption of the circadian system is evident in PD [27] and associated with cognitive impairment [28]. Although we did not measure functionality of the circadian system, the presented data could help to better understand this association. Patients with α-synucleinopathies may have difficulties entraining to the light-dark cycle due to disruption of ipRGCs, with a detrimental impact on cognition. It would be important to examine whether existing treatment strategies aimed at stabilizing the circadian system (e.g., through light [29] or melatonin [30, 31]) can improve cognition in α-synucleinopathies and might positively impact melanopsin functioning. Longitudinal comparisons of the PIPR before and after such interventions would be of great interest.
In conclusion, data show an association between specific subcortical cognitive performance levels and the PIPR in patients with iRBD. Since chromatic pupillometry is a robust, easy-to-use, non-invasive, and inexpensive method, the PIPR may prove a valuable functional biomarker to indicate impaired cognitive performance levels in the advanced state of prodromal α-synucleinopathies. Moreover, PIPR holds promise to be therapy responsive and thus would be a monitoring biomarker.
Footnotes
ACKNOWLEDGMENTS
We would like to thank Dr. Michael Ehrensperger (Deputy Head Memory Clinic Basel, Switzerland) for his advice on how to calculate the two cognition composite scores. We would also like to thank Dr. Lorenz Grolig and Sebastian Fischer for their assistance during data collection.
CONFLICT OF INTEREST
The authors have no conflict of interest to report.