Regional Cerebral Cholinergic Vesicular Transporter Correlates of Visual Contrast Sensitivity in Parkinson’s Disease: Implications for Visual and Cognitive Function

Abstract

Visual and visual processing deficits are implicated in freezing, falling, and cognitive impairments in Parkinson’s disease (PD). For example, contrast sensitivity deficits are common and may be related to cognitive impairment in PD. While dopaminergic deficits play a role in PD-related visual dysfunction, brain cholinergic systems also modulate many aspects of visual processing. The aim of this study was to explore regional cerebral cholinergic terminal density correlates of contrast sensitivity in PD. Ninety-one PD subjects underwent contrast sensitivity testing, motor testing, cognitive testing, and brain MRI and [¹⁸F]-fluoroethoxybenzovesamicol [¹⁸F]-FEOBV PET imaging. Whole brain false discovery error-corrected (p < 0.05) correlations revealed significant associations between VAChT deficits in pericentral, limbic, and visual processing regions and contrast sensitivity performance, independent of disease duration and dopaminergic medication doses. These results suggest that brain cholinergic deficits correlate with contrast sensitivity deficits in PD. Additionally, decreased Rabin contrast sensitivity scores were associated with lower total scores in the Parkinson’s Disease Cognitive Rating Scale. These findings suggest that diminished cognitive performance correlated with contrast sensitivity partly reflects underlying vulnerabilities of brain cholinergic systems.

1. Introduction

Individuals with Parkinson’s disease (PD) experience an array of motor and non-motor symptoms impacting daily life functions [1]. While the impact and mechanisms of motor features are a traditional focus of PD research, non-motor features are increasingly recognized as an important determinant of disability in PD. Visual disturbances are a frequently reported non-motor feature and are seen at a higher rate in older adults with PD than elderly controls [2]. Visual disturbances in PD may be associated with increased falls risk, cognitive decline, and postural instability and gait disturbances (PIGD) [2,3]. Changes in retinal function have been the focus of investigating visual dysfunction in PD, but brain visual system processing changes may contribute to visuospatial dysfunction as well as deficits in visual acuity and contrast sensitivity [3].

Diminished contrast sensitivity, which affects the ability to distinguish objects from backgrounds, heightens the risk of functional limitations and poses safety hazards, such as increased likelihood of falls and fall-related injuries. PD subjects exhibit lower scores on visual contrast sensitivity tests, both in high and low contrast settings [4–6]. Contrast sensitivity impairments in PD have been associated with a combination of retinal dopaminergic deficits and changes in cortical visual processing regions [3,7–9]. While dopaminergic deficits are well documented in contributing to visual dysfunction in PD, other neurotransmitter systems, such as acetylcholine, may also play a role. The retina contains cholinergic interneurons and numerous components of the central visual system receive cholinergic inputs from cholinergic basal forebrain, pedunculopontine-lateral dorsotegmental, and parabigeminal neurons. Recent studies demonstrated cholinergic modulation of responses to visual stimuli, including increasing contrast sensitivity [10–12]. Bhattacharyya et al. [13] demonstrated that basal forebrain activation has a strong influence on primary visual cortex contrast sensitivity through both cholinergic and GABAergic projections. These studies raise the possibility that brain cholinergic system deficits contribute to impaired contrast sensitivity in PD.

Impaired contrast sensitivity is associated with other non-motor features of PD, including cognitive impairments, other visual disturbances, olfactory deficits, sleep disturbances, and mood disorders [4–7,14–16]. In a longitudinal study, Hong et al. [4] described an association between contrast sensitivity and cognitive decline in drug-naïve PD participants, demonstrating lower contrast sensitivity at baseline correlated with poorer performance on the Mini-Mental State Examination (MMSE) at baseline and follow-up. Visual pathologies in PD often coincide with higher-order processing tasks including visuospatial skills, working memory, navigation, and spatial perception [17–20]Previous work from our group showed a significant correlation between mean contrast sensitivity scores and cognitive domain performance, including executive function, verbal learning, visuospatial skills, and attention [7]. Recent research has demonstrated the role of cholinergic deficits and impairments within specific cognitive domains in PD [15]. These cumulative findings suggest a close relationship between cholinergic vulnerability, visual impairments, particularly contrast sensitivity, and several cognitive functions in PD. The primary objective of this study was to explore correlations between regional cerebral cholinergic terminal density and contrast sensitivity using [¹⁸F]-fluoroethoxybenzovesamicol ([¹⁸F]-FEOBV) VAChT PET in PD subjects. Our overall hypothesis was that poorer performance on the Rabin contrast sensitivity test would be significantly correlated with cholinergic denervation. Additionally, we hypothesized that cholinergic denervation in regions associated with the Rabin contrast sensitivity test scores would be associated with cognitive deficits.

2. Materials and Methods

2.1. Protocol approval and subject consent

This study was approved by the Institutional Review Boards of the University of Michigan and Ann Arbor Veterans Affairs Healthcare System. All participants provided written informed consent prior to their participation. This research was supported by the NIH, Michael J. Fox Foundation, and the Parkinson’s Disease Foundation [grant numbers P50 NS123067, P01 NS015655, P50 NS091856, and PF-RCE-1947].

2.2. Participants

This retrospective cross-sectional study included 91 subjects with PD (67 males, 24 females, mean age of 66.4 ± 6.3). We retrieved data from PD patients who underwent the Rabin contrast sensitivity tests, Parkinson’s Disease Cognitive Rating Scale (PD-CRS), [¹⁸F]-FEOBV PET scans, and MRI from our database in our center. Subjects met the United Kingdom Parkinson’s Disease Society Brain Bank Research Centre clinical diagnostic criteria for PD [21]. Typical striatal dopaminergic denervation was confirmed in all PD participants by [¹¹C]-dihydrotetrabenazine (DTBZ) vesicular monoaminergic type 2 transporter PET or by cocaine analog N-(3-iodoprop-2E-enyl)-2β-carbomethoxy-3β-(4-methyl-phenyl)nortropane [¹¹C]-PE2I PET. Subjects with evidence of large vessel stroke or other intracranial lesions on anatomic imaging were excluded. Subjects with ophthalmologic conditions, including glaucoma, cataracts, macular degeneration, retinal neuropathy, astigmatism, and double vision, were excluded. Disease staging was assessed by categorizing PD patients into two groups: early stage and mid-advanced stage. The early stage was defined as having motor symptoms for five years or less. REM sleep behavior disorder (RBD) assessed through self-report, with participants indicating “yes” to the question of whether they act out their dreams. Utilizing the MDS-UPDRS and the criteria outlined by Stebbins et al. [22] participants were classified into PIGD and tremor-dominants (TD) subtypes.

2.3. Clinical Assessments

Clinical evaluations of PD subjects included a standardized neurological assessment, demographic information, medication history, and assessment of overall PD severity using the Movement Disorder Society-revised Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). Motor fluctuations were assessed using item 4.3 from the MDS-UPDRS-IV (“Time spent in OFF state”), classifying those with a score of 1 or higher (“≤ 25% of waking day”) as having motor fluctuations. Visual acuity was assessed using the Snellen eye chart. Cognitive function was evaluated using the PD-CRS, with lower scores indicating worse cognitive function. The PD-CRS is broken up into nine subtests: 1) immediate free recall verbal memory; 2) confrontation naming; 3) sustained attention; 4) working memory; 5) unprompted drawing of a clock; 6) copy drawing of a clock; 7) delayed free recall verbal memory, 8) alternating verbal fluency; 9) action verbal fluency. Visual contrast sensitivity was assessed using the Rabin Contrast Sensitivity Test. Subjects were tested in a dark room 4 meters from an illuminated light cabinet (Precision Vision^®, La Salle, IL, USA). All subjects completed the test “on” their dopaminergic medication. The light box included eight rows of five letters with each row decreasing in contrast. Subjects were instructed to occlude one eye and read row-by row down the chart and then repeat with the opposite eye occluded. The test administrator totaled the number of letters missed for each eye and bilaterally averaged the errors. Contrast sensitivity was then scored on a scale from 0 to 2, in increments of 0.05, with lower scores indicating worse sensitivity performance.

2.4. Imaging acquisition and pre-processing

Brain MRI was conducted using a 3 Tesla Philips Achieva system (Philips, Best, The Netherlands), while PET imaging was performed in 3D imaging mode with an ECAT Exact HR + tomograph (Siemens Molecular Imaging, Inc., Knoxville, TN) [23] or a Biograph 6 TruPoint PET/CT scanner (Siemens Molecular Imaging, Inc., Knoxville, TN) as previously reported [24]. To ensure unbiased results, an inter-scanner normalization method was implemented, along with corrections for scatter and motion [25]. The preparation of the [¹⁸F]-FEOBV was previously described by Shao et al. [26]. Delayed dynamic imaging of the [¹⁸F]-FEOBV was performed over 30 minutes (in six 5-minute frames) starting 3 hours after an 8 mCi intravenous bolus dose injection of the tracer [27]. MRI and PET images were co-registered for each subject using MRI-PET registration statistical parametric mapping (SPM) software (SPM12; Welcome Trust Centre for Neuroimaging, University College, London, England [ https://www.fil.ion.ucl.ac.uk/spm/software/spm12/]). Parametric images for the [¹⁸F]-FEOBV PET were created using a supratentorial white matter reference tissue approach as previously reported [28, 29]. Freesurfer was utilized for white matter reference regions. We excluded the white matter voxels below and lateral to the ventricles and boundary voxels with the cortical areas. To reduce motion artifacts, PET frames were aligned utilizing rigid-body transformation within each subject. We applied SPM-based segmentation of MR images into gray matter, white matter, and cerebrospinal fluid to undergo correction for partial volume effects (PVE) in PET images using the Müller-Gartner method [30]. The PVE-corrected parametric PET images and the registered structural MR images underwent further processing using high-dimensional DARTEL registration and spatial normalization to a template space in the Montreal Neurological Institute (MNI) space utilizing a full width at half maximum of 8 mm to reduce random noises as previously described [25].

2.5. Statistical Analyses

The primary objective of this study was to determine if regional cerebral [¹⁸F]-FEOBV uptake is correlated with changes in contrast sensitivity in PD. Whole-brain voxel-based [¹⁸F]-FEOBV PET and Rabin scores correlational analyses were performed using parametric SPM12 software after adjustment for levodopa equivalent dose (LED) and disease duration. For both the whole brain voxel-based and the peak cluster analysis significance thresholds after voxel-level false discovery rate (FDR) correction was set at p < 0.05 cluster extent (K)> 50 voxels.

Additionally, we aimed to integrate cognition into our analysis by exploring the relationship between contrast sensitivity, cholinergic deficits, and cognition through a series of linear regression models: i) the association between contrast sensitivity and cognition while accounting for disease duration and LED levels; and ii) the association between contrast sensitivity-related cholinergic substrate and cognition. For this analysis, we extracted DVR values for each patient using the significant clusters from the voxel-wise SPM12 correlation analyses as a VOI. These VOIs were used to extract DVR values from participants’ [¹⁸F]-FEOBV scans, deriving mean [¹⁸F]-FEOBV binding values for regions significantly associated with Rabin scores. Prior to including PD-CRS total and sub-scores in the regression models, we conducted exploratory analyses to ensure sufficient variability and to examine the data distribution. If a variable exhibited extreme skewness, with most observations falling under a single value, indicating a ceiling effect, it was excluded from subsequent analyses due to the lack of variability, preventing reliable estimation of any association. The normality of the regression residuals was assessed using the Shapiro-Wilk test for all the regression models. If the residuals were found to violate the assumption of normality, we ran the model using robust regression, which does not require this assumption. All the analyses described above were performed using R Studio.

To further investigate the association between Rabin scores, cognition, and the cholinergic system, we conducted mediation analyses to examine i) whether the Rabin score (X) was independently associated with the cholinergic system (Y) regardless of cognition (mediator M), and ii) whether the association between Rabin scores (X) and cognition (Y) was mediated by the common cholinergic substrate (M) or independent of cholinergic innervation. The confidence intervals were estimated with 10,000 bootstrapped samples. Mediation analyses were run with JASP Team JASP Team (2024). JASP (Version 0.18.3) [Computer software]. Retrieved from https://jasp-stats.org/. Lastly, we conducted a series of post-hoc sensitivity analyses to investigate specific confounding factors, including age, visual acuity, visual hallucinations, disease duration, motor fluctuations, RBD and motor subtypes (TD and PIGD). (See supplementary material S1 for details).

3. Results

3.1. Clinical and Demographic information

Subjects had mild to moderate disease severity with a mean modified Hoehn and Yahr [31,32] stage of 2.4 ± 0.6 and a mean disease duration of 5.89 ± 4.6 years. Subjects had an average levodopa equivalence dosage of 597.66± 407.64 mg. Four participants reported visual hallucinations. Forty-seven participants experienced RBD. Thirty-five participants met the criteria for PIGD, 47 met the criteria for TD, and and9 were indeterminate. Participants had a mean PD-CRS score of 91.10 ± 9.84, with 14.3% of participants meeting the criteria for mild cognitive impairment (score of < 81) [33]. For visual assessments, participants had a mean score of 29.41± 14.19 on the best corrected Snellen visual acuity test and a mean log score of 1.52 ± 0.28 on the Rabin Contrast Sensitivity Test. For more comprehensive demographic details, refer to Table S1.

3.2. Whole brain [¹⁸F]-FEOBV PET voxel-based correlations with Rabin contrast sensitivity scores

The whole-brain voxel-based correlation analysis (FDR-corrected, p < 0.05) revealed significant correlations between [¹⁸F]-FEOBV binding and total Rabin scores across specific brain clusters. These correlations encompassed bilateral brain regions, including frontotemporal areas such as the anterior-mid-posterior cingulate, Rolandic operculum, inferior-middle-superior frontal gyrus, precentral gyrus, postcentral gyrus, paracentral lobule, supplementary motor area, inferior-middle-superior temporal gyrus, fusiform gyrus, Heschl’s gyrus, inferior-middle-superior temporal gyrus, and inferior temporal gyrus. Additional parietal correlations included the inferior-middle-superior parietal gyrus, angular gyrus, and supramarginal gyrus. Occipital regions included bilateral calcarine, cuneus, lingual gyrus, right more than left posterior orbital gyrus, and middle-superior occipital gyrus. Limbic regions involve the bilateral hippocampus, insula, and para-hippocampal gyrus (Figure 1 and Table S2).

Figure 1.

SPM voxel-based analysis of Rabin scores and [¹⁸F]-FEOBV (FDR p < 0.05, K > 50). The figure follows the neurological convention where right is represented as the right side and left as the left side.

3.3. Contrast Sensitivity and Cognitive Test Associations

Exploratory analyses on PD-CRS revealed that the total scores, immediate and delayed verbal memory recall, as well as alternating and action verbal fluency were normally distributed. However, sustained attention, spontaneous clock drawing, copy of the clock, working memory, and confrontation naming were abnormally distributed. Among the non-normally distributed variables, sustained attention (82.42% of patients scored either 9 or 10, the maximum), spontaneous clock drawing (87.91% of patients scored either 9 or 10, the maximum), and copy of the clock (95.60% of patients scored either 9 or 10, the maximum) exhibited a ceiling effect, indicating insufficient variability. Consequently, these variables were excluded from subsequent analyses (Figure S1).

Linear regression revealed significant associations between Rabin scores and total PD-CRS scores after controlling for disease duration and LED. The overall regression was statistically significant, where lower Rabin scores were associated with lower PD-CRS total scores (t(87) = 3.957, p < 0.001). Significant associations were also seen in the models for immediate free recall (t(87) = 2.554, p = 0.012), delayed free recall (t(87) = 2.347, p = 0.021), working memory (t(87) = 2.900, p = 0.005), alternating verbal fluency (t(87) = 2.626, p = 0.010), and action verbal fluency (t(87) = 2.069, p = 0.039). No significant associations were seen for confrontation naming (t(87) = 0.523, p = 0.602). Bonferroni corrections were applied to correct for multiple comparisons. After corrections, both working memory (p_bonf = 0.033) and total PD-CRS scores (p_bonf < 0.001) remained significant. See Table 1 for detailed regression parameters.

Table 1.

Linear regression between Rabin contrast sensitivity and PD-CRS sub and total Scores.

Model	B	t-value	p-value	Adjusted p-value	95% CI		Adj R2
					Lower	Upper
Confrontation Naming*	0.072	0.523	0.602	1.000	−0.200	0.343	−0.015
Working Memory*	0.264	2.900	0.005	0.033	0.083	0.445	0.056
Immediate Free Recall Verbal Memory*	0.245	2.554	0.012	0.087	0.054	0.436	0.051
Delayed Free Recall Verbal Memory*	0.270	2.347	0.021	0.148	0.041	0.498	0.060
Alternating Verbal Fluency*	0.276	2.626	0.010	0.071	0.067	0.484	0.060
Action Verbal Fluency*	0.189	2.069	0.039	0.273	0.010	0.368	0.021
Total PD-CRS*	0.355	3.957	0.0002	0.001	0.177	0.534	0.120

3.4. Mean Uptake of Significant [¹⁸F]-FEOBV Cluster Voxel Binding and Linear Regression

We observed that lower mean uptake of the significant voxel-[¹⁸F]-FEOBV binding, extracted from the clusters (FDR p < 0.05) associated with lower Rabin scores, was significantly associated with lower total PD-CRS total scores (t(89) = 3.500, p < 0.001). Significant associations were seen in PD-CRS sub-scores with immediate free recall (t(89) = 3.968, p < 0.001) and delayed free recall (t(89) = 3.694, p < 0.001). No significant results were seen for action verbal fluency (t(89) = 1.107, p = 0.271), confrontation naming (t(89) = 0.959, p = 0.340), alternating verbal fluency (t(89) = 1.397, p = 0.166) and working memory (t(89) = 1.560, p = 0.122). Bonferroni corrections were applied to correct for multiple comparisons. After the correction, immediate free recall (p_bonf < 0.001) delayed free recall (p_bonf = 0.003), and total PD-CRS scores (p_bonf = 0.005) remained significant. See Table 2 for detailed regression parameters.

Table 2.

Linear regression between significant mean uptake of [¹⁸F]-FEOBV binding and PD-CRS sub and total scores.

Model	B	t-value	p-value	Adjusted p-value	95% CI		Adj R2
					Lower	Upper
Confrontation Naming	0.101	0.959	0.340	1.000	−0.108	0.311	0.013
Working Memory	0.168	1.560	0.122	0.857	−0.046	0.383	0.030
Immediate Free Recall Verbal Memory	0.386	3.968	0.0002	0.001	0.193	0.579	0.155
Delayed Free Recall Verbal Memory	0.364	3.694	0.0004	0.003	0.168	0.560	0.138
Alternating Verbal Fluency	0.149	1.397	0.166	1.000	−0.063	0.361	0.024
Action Verbal Fluency	0.118	1.107	0.271	1.000	−0.094	0.331	0.016
Total PD-CRS	0.354	3.500	0.0007	0.005	0.153	0.554	0.126

3.5. Mediation Analysis between Rabin Scores, PD-CRS Total, and the Mean Uptake of Significant [¹⁸F]-FEOBV Cluster Voxel Binding.

The total PD-CRS score was the only variable that remained significantly associated with both [¹⁸F]-FEOBV DVR and Rabin contrast sensitivity scores after Bonferroni corrections. Therefore, it was included in the mediation analysis. The direct association between Rabin scores and [¹⁸F]-FEOBV binding was significant, independent of PD-CRS scores (direct effect: β = 1.29, SE = 0.353, p < 0.001). PD-CRS scores also mediated this relationship (indirect effect: β = 0.314, SE = 0.153, p = 0.040; total effect: β = 1.60, SE = 0.340, p < 0.001) (Figure 2A). Additionally, the direct association between Rabin scores and total PD-CRS scores was significant (direct effect β = 0.88, SE = 0.382, p = 0.021), when the mean uptake of significant [¹⁸F]-FEOBV cluster voxel binding was used as the mediator. The [¹⁸F]-FEOBV binding mediated the association (indirect effect β = 0.418, SE = 0.191, p = 0.029; total effect: β = 1.302, SE = 0.354, p < 0.001) (Figure 2B).

Figure 2.

Mediation analysis between total PD-CRS scores, Rabin scores, and mean [¹⁸F]-FEOBV uptake.

3.6. Post-Hoc sensitivity Analyses

Age -

The age range for the participants ranged from 52–82. To assess potential age-related effects of age on Rabin scores, we ran a regression analysis with mean uptake of significant [¹⁸F]-FEOBV cluster voxel binding with Rabin scores and age. Rabin contrast sensitivity scores were associated with DVR values of the significant cluster (β = 0.38, CI 95% [0.20: 0.56], p < 0.001), independent of age (β = −0.37, CI 95% [−0.55: −0.19], p < 0.001). There was no significant interaction effect between the two variables in this association (β = −0.02, CI 95% [−0.19: 0.22], p = 0.872). The residuals of the model were normally distributed, and the model explained 31% of the variance (Adjusted R-squared = 0.30, p < 0.001). This suggests that while age is a factor influencing cholinergic innervation status, Rabin scores independently contribute to the innervation of these specific structures (Figure S2).

Visual acuity -

We conducted a whole-brain voxel-based correlation analysis of [¹⁸F]-FEOBV binding with the Snellen visual acuity scores under the “both eyes open” condition and found no significant correlations.

Visual Hallucinations –

PD patients with (n = 4) and without (n = 87) visual hallucinations did not show a significant difference in the distribution of Rabin contrast sensitivity log scores, but this analysis was limited due to the small numbers of persons with hallucinations (Figure S3)

Disease duration and motor fluctuations -

Patients with motor fluctuations were evenly distributed across the early and mid-late stages within the subset of patients with complete information available (Table S3). Early-stage PD patients (n = 50) and mid-advanced PD patients (n=41) did not differ significantly in any variable except for LED, where the mid-advanced group showed significantly higher values than the early-stage group (p < 0.001) (Table S4). Patients with motor fluctuations (n =17) and those without (n = 44) did not show any statistically significant differences (Table S5).

RBD-

We compared the Rabin contrast sensitivity log scores and PD-CRS total scores between individuals who reported acting out their dreams (RBD) and those who did not using a Welch two sample t-test. One participant did not have data for RBD and was removed from the analysis. PD patients with (n = 47) and without (n = 43) RBD did not show any significant differences in the distribution of Rabin contrast sensitivity log scores (p = 0.80) and total PD-CRS scores (p = 0.67) (Figure S4–S5).

Motor subtypes -

PD patients with TD and PIGD subtypes did not differ significantly in Rabin contrast sensitivity log scores (p = 0.7888) (Figure S6).

4. Discussion

In this study, we investigated cerebral cholinergic deficits associating with contrast sensitivity, as assessed by the Rabin contrast sensitivity test, in individuals with PD. Using the specific radiotracer [¹⁸F]-FEOBV, which targets cholinergic terminals, we found significant correlations between lower regional [¹⁸F]-FEOBV binding and poorer performance on the Rabin contrast sensitivity test, independent of effects from dopaminergic medication and disease duration. Our post-hoc sensitivity and mediation analyses revealed that this association remains unaffected by cognitive impairment (only partially mediated by it), as well as by age, visual acuity, visual hallucinations and motor fluctuations. These findings suggest that cholinergic degeneration may serve as a pathological substrate for impairments in contrast sensitivity in PD.

Specifically, we observed robust cholinergic deficits linked to performance on the Rabin contrast sensitivity test in various brain regions, including the pericentral, limbic, paralimbic, and occipital areas. We also examined the relationship between contrast sensitivity and cognition using the PD-CRS. Our analysis showed that Rabin scores were significantly associated with total PD-CRS scores and specifically with the working memory item. Moreover, lower mean [¹⁸F]-FEOBV binding from the significant clusters (FDR p < 0.05) significantly predicted total PD-CRS scores, along with immediate and delayed free recall verbal memory sub-scores. The mediation analysis demonstrated that cholinergic vulnerability partially mediated the association between Rabin scores and cognitive function, indicating a shared cholinergic deficit in the pericentral, limbic, and visual processing regions between contrast sensitivity and cognition in PD. These findings are consistent with prior observations that dysfunctional cholinergic networks in limbic-paralimbic regions are associated with cognitive deficits, including dementia, in PD [20, 34]. Cholinergic dysfunction in (para)limbic regions, specifically the insula, has been linked to visual dysfunction in PD [20, 35]. In addition to limbic regions, our findings demonstrate correlated cholinergic denervation in pericentral regions, including parts of the centro-cingulate network, which overlaps with the cingulo-opercular task control network (COTC), an important modulator of cognition and alertness [36]. Recent research shows that denervation of the centro-cingulate network correlates with cognitive deficits in domains including memory, executive function, attention, and language [15]. In a recent study by Bohnen et al. [36], key areas of the centro-cingulate network, such as the posterior cingulum, insula, and the retrosplenial cingulum, all correlated with memory dysfunction. Our study highlights the vulnerability of cholinergic systems in regions associated with reduced contrast sensitivity and memory deficits, specifically the insula, peri-central cortices, posterior cingulum, and retrosplenial cingulum.

Extending the work of Ridder et al. [7], our research suggests an intersection between verbal memory and contrast sensitivity, potentially linked to disrupted information transmission from the hippocampus to the posterior cingulate, a key hub for multimodal visual processing within the cortex [37]. These findings suggest that contrast sensitivity is closely related to spatial navigation, the centro-cingulate network, and the ventral attention network [38,39]. The limbic and paralimbic deficits linked to contrast sensitivity impairment may underscore the significance of the cholinergic system in integrating sensory information, including spatial navigation, potentially indicating impairments in top-down processing [40–42].

Other visual impairments such as glaucoma, cataracts, hallucinations and double vision may also impact contrast sensitivity, all of which are common in an older population and those with PD [43,44]. While participants with visual impairments were excluded from the study, four participants self-reported having visual hallucinations. Participants with and without visual hallucinations had overlapping Rabin scores, suggesting no significant difference in contrast sensitivity between the groups; however, the small sample size limits meaningful effects (Figure S3). Due to the retrospective design of the study, we were unable to assess asymmetry in contrast sensitivity or determine whether the decreased sensitivity resulted from a unilateral or bilateral deficit. While we believe that using the log contrast scores of both eyes accurately reflects contrast sensitivity in everyday activities, future studies should investigate the potential effects of ocular asymmetry in contrast sensitivity in PD. Dopaminergic losses in the retina and inner retinal thickness may also be implicated in decreased contrast sensitivity scores [8,44]. Participants were tested while on dopaminergic medication, which may influence overall contrast sensitivity scores [45]. To mitigate this influence, we accounted for LED scores in our analyses. Attention, visuospatial and visuo-constructive domains were excluded from our analyses due to limited variability in the dataset. Drawing on prior research from our group, Ridder et al. [7], we hypothesize attention and visuospatial correlations with both contrast sensitivity scores and cholinergic vulnerability, although these associations could not be explored in the current study. Due to the cross-sectional and retrospective design of this study, we are also not able to see how changes in contrast sensitivity may predict cognitive decline or conversion to dementia. Another limitation is that we were not able to evaluate the integrity of retinal cholinergic neurons. The existence of both central and retinal dopaminergic neuron degeneration in PD raises the possibility that this may be true for retinal cholinergic neurons. Evaluation of the integrity of retinal cholinergic function in PD may be fruitful.

These findings suggest cholinergic denervation in regions important for cognition is also independently associated with contrast sensitivity test performance. The brain voxel-based analysis combined with the linear regression shows promising evidence of the connection between decreased contrast sensitivity, associated cholinergic deficits, and cognitive impairments in PD. This aligns with recent findings where uncorrected visual dysfunction may be a new risk factor for dementia [46]. Our findings suggest these correlations may be due in part through shared underlying cholinergic deficits, at least in PD. Furthermore, the relationship with other factors, such as comorbid diabetes and RBD should also be explored [46]. We studied the presence of RBD in our cohort and its association with contrast sensitivity but found no significant difference. This lack of difference may be attributed to the self-reported nature of RBD, as it is plausible that some PD patients who do not report RBD may experience it without being aware, due to reasons such as never having fallen out of bed, not having a bed partner, or experiencing only mild symptoms. Our research expands upon previous studies by revealing that diminished cognitive performance associated with contrast sensitivity may not solely result from retinal dopamine loss, but also from underlying vulnerabilities of brain cholinergic systems in PD.