An exploratory study to investigate the utility of clinical screening for neurodegenerative disease in age-related eye disease research

Age-related neurodegenerative diseases (NDD) that lead to dementia, such as Alzheimer’s disease, Parkinson’s disease, and dementia with Lewy bodies, result in retinal degeneration and brain-related visual dysfunction^1–4. These diseases have in common long pre-dementia stages, which include a preclinical stage and a mild cognitive impairment (MCI) stage, and affected individuals can harbor pathological brain changes for at least two decades and retinal changes for up to one decade or more prior to dementia onset^5–7. In the late 1980s, Sadun and colleagues found evidence for loss and degeneration of retinal ganglion cell bodies and axons in post-mortem histopathological studies of retinas from persons with neuropathological evidence of Alzheimer’s disease but not in age-matched controls². Over the past two decades, retinal optical coherence tomography (OCT) of the macula and the circumpapillary regions reveal that the inner retinal loss and macular retinal volume and thickness loss, can be demonstrated during life in the dementia and MCI stages and potentially in preclinical stages^3,6.

These findings are consequential to age-related eye disease studies, because NDD are common and people in the early stages are often not recognized or diagnosed until the late stages of dementia⁸. For instance, it is estimated that 28-38% of people > 65 years are in the pre-dementia or dementia stages of NDD, and only 20-50% of cases of dementia are diagnosed and documented by primary care providers, with a much smaller proportion identified in the pre-dementia stages^8,9. NDD manifestations are frequently mistaken for normal aging, although thorough neurologic and cognitive evaluations and retinal OCT can detect clinicopathological changes^2,5,10.

Cortical and subcortical visual pathways are also selectively vulnerable to NDD, and higher order visual brain dysfunction occurs while normal visual acuity is maintained, and functional ability is strongly correlated with visuospatial function and semantic memory more than any other cognitive domain in early Alzheimer’s disease¹¹. Vision-specific quality of life measures, such as the National Eye Institute Visual Functional Questionnaire – 25 (VFQ-25), assess everyday activities that depend on higher order visual processing (e.g., reading and driving) and are compromised in people with Parkinson’s disease and other parkinsonian syndromes¹². These data suggest that undiagnosed NDD could impact vision-specific quality of life measures used in age-related eye disease research, but this has not been previously investigated.

It is unknown whether screening for undiagnosed neurodegenerative diseases or cognitive impairment as a potential confounder in age-related eye disease research is necessary. Thus, the objective of this cross-sectional, single center exploratory study was to investigate the relationship between NDD screening tools and VFQ-25 and retinal OCT measures in healthy controls without known NDD or cognitive impairment in an age-related macular degeneration registry, and to use these data to determine the utility of larger, follow up studies.

Methods

Participants.

After approval from the Colorado Multiple Institutional Review Board, healthy controls from the Colorado Age-Related Macular Degeneration (CO-AMD) Registry at the University of Colorado, described in detail elsewhere¹³, were screened and those meeting inclusion and exclusion criteria were invited to participate. Inclusion criteria included enrollment in the CO-AMD Registry as a control subject (cataract surgery patients without AMD) and age between 60 to 75 years. Exclusion criteria included dense cataract precluding optical imaging of the retina, decrease in visual acuity due to preexisting retinal disease, history of panretinal photocoagulation for diabetic retinopathy, history of branch or central retinal vein occlusion, prior ocular inflammatory disease, moderate or severe glaucoma or other known optic neuropathy, active cancer, previous diagnosis of NDD, known history of cognitive impairment or dementia, poor health or gravely ill, or status as a prisoner or on probation. None of the participants had cognitive complaints. Participants were recruited between October 2019 and February 2020. Study data were collected and managed using REDCap (Research Electronic Data Capture) tools.

Macular OCT Imaging.

All participants underwent spectral-domain optical coherence tomography (SD-OCT) using the Heidelberg Spectralis, software version 6.6.2 (Heidelberg Engineering GmbH, Heidelberg, Germany). Right and left eye macular SD-OCT scanning with scan fixation on the fovea was performed by trained ophthalmic technicians at the UCHealth Sue Anschutz-Rodgers Eye Center (Department of Ophthalmology, University of Colorado School of Medicine). Automated segmentation techniques were used for macular retinal ganglion cell layer (RGCL) thickness and volume measures. Manual segmentation of the macular retinal nerve fiber layer (RNFL) was performed by one author (Y.M.P). Post-scan quality control included review of all retinal images by a vitreoretinal ophthalmologist, and segmentation and signal quality review by an ophthalmology-trained neuro-ophthalmologist and a neurology-trained neuro-ophthalmologist (P.S.S and V.S.P). Quality control reviewers were blinded to NDD screening results.

Neurological and Cognitive Screening.

Participants had one study visit with the following procedures performed by a board-certified neurologist and neuro-ophthalmology fellow (Y.M.P): Neurologic and neuro-ophthalmic examinations; the VFQ-25 (completed at the visit if the initial VFQ-25 from the CO-AMD Registry was older than 6 months or if cataract extraction occurred after the initial VFQ-25); the Montreal Cognitive Assessment (MoCA) Version 7.0; the Colorado Parkinsonian Checklist (CPC – described below); and the Lewy Body Composite Risk Score (LBCRS)¹⁴. Clinically relevant findings (e.g., abnormal examination or abnormal MoCA scores) were shared with participants, who were given the option for findings to be shared with their primary care provider. The MoCA was chosen since it was designed to capture mild cognitive impairment (MCI) and is more sensitive and specific than the Mini-Mental Status Examination for MCI and Alzheimer’s disease¹⁵. It has a positive predictive value of 98% and a negative predictive value of 89% for Alzheimer’s disease, and scores have empirical validation compared to the gold standard, formal neuropsychological testing¹⁶. A score of ≥ 26 of 30 is considered normal¹⁵. The Lewy Body Composite Risk Score (see Table, Supplemental Digital Content 1) has a maximum score of 10 points and three or more points is considered high risk for Lewy body dementia¹⁴. For this study, we created a Colorado Parkinsonian Checklist (see Table, Supplemental Digital Content 2) (maximum 11 points) to pilot the checklist as a method to standardize the assessment for features of Parkinson’s disease (PD) and parkinsonism and to enumerate diagnostic criteria from the Movement Disorder Society’s diagnostic criteria for PD¹⁷.

Statistical Analysis:

Descriptive statistics included basic frequencies for categorical variables, and means, standard deviations, medians and ranges were used for continuous variables. Univariate and multivariable linear regression modeling was used to assess associations between the explanatory variables that included the neurologic and cognitive screening measures (i.e., MoCA, LBCRS, CPC) and the following outcome variables: VFQ-25; VFQ-25 subscales (for driving, near activities, peripheral vision) and VFQ-25 question #7, because it assesses the ease of finding objects on a crowded shelf, which is commonly reported by those with posterior cortical dysfunction; and macular OCT measures were chosen based on previously established relationships between macular OCT and NDD, including macular OCT volume for RNFL and RGCL, and macular OCT thicknesses for RNFL and RGCL using the ETDRS (Early Treatment of Diabetic Retinopathy Study) OCT macular grid.

Results

Twenty-nine participants were recruited, and one patient withdrew due to discomfort with cognitive testing. Twenty-four right eye and 20 left eye macular OCT scans were eligible for analysis, and OCT studies were completed within an average of 17 months prior to neurologic and cognitive screening. There was agreement by retina and neuro-ophthalmology experts regarding the quality review of OCT measures, which eliminated four right eye scans and eight left eye scans due to lack of complete data for the outer ring or evidence of asymptomatic macular traction, and in one eye, evidence of an asymptomatic macular hole.

The average age for participants was 72.8±3.9 years with 19 females (67.9%) (Table 1). The mean MoCA score was 26.5 (median 28; range 20-30). Criterion for cognitive impairment (MoCA score <26/30) was met for 10/28 (36%) participants and none had MoCA scores within the dementia range (i.e., <20). Examples of incorrect figure copy and clock drawing, completed for the MoCA by participants scoring <26/30, are shown in Figure 1. Four participants had a score of one on the LBCRS and all others had a score of zero. Similarly, three participants had a score of one on the CPC (and each with one point on the LBCRS), and all others had a score of zero on the CPC. LBCRS screening noted REM sleep behavior disorder (n = 1), orthostasis (n = 1), bradykinesia (n = 1), and excessive daytime drowsiness (n = 1). Of note, the participant with bradykinesia scored one point for small handwriting on the CPC questionnaire; the other two participants scored one on the CPC for orthostasis and one for REM sleep behavior disorder, overlapping with the LBCRS. The mean total score for the VFQ-25 for all participants was 95.2±4.3 (Table 1). The average macular OCT measurements for right eyes and left eyes are shown in Table 2.

Table 1.

Demographic factors and questionnaire summary measures, total = 28 participants.

	n (%)
Demographics
Gender
Male	9 (32.1%)
Female	19 (67.9%)
Age (years)
Mean (SD)	72.8 (3.9)
NDD Screening Tests
Montreal Cognitive Assessment
Mean (SD)	26.5 (3.0)
Median (range)	28.0 (20-30)
Colorado Parkinsonism Sign and Symptom Checklist
Mean (SD)	0.11 (0.31)
Median (range)	0 (0-1)
Score of 1	3 (10.7%)
Lewy Body Composite Risk Score
Mean (SD)	0.14 (0.36)
Median (range)	0 (0-1)
Score of 1	4 (14.3%)
NEI Visual Function Questionnaire-25
VFQ-25 composite score
Mean (SD)	95.2 (4.3)
Median (range)	96.9 (80.8-100)

Figure 1.

Drawings from four participants. Clock drawings (from the Montreal Cognitive Assessment or MoCA) by two participants with MoCA scores of 24 points (A) and 20 points (B) out of a total possible 30 points. Cube Copy test drawings (from the MoCA) by two other participants with MoCA scores of 24 (C) and 20 (D).

Table 2.

Spectral domain - optical coherence tomography measures for right and left eye macula.

	Average Macular Right Eye RNFL N Mean (SD)	Average Macular Left Eye RNFL N Mean (SD)	Average Macular Right Eye RGCL N Mean (SD)	Average Macular Left Eye RGCL N Mean (SD)
OCT volume (mm3)	20 1.0 (0.13)	18 1.0 (0.12)	20 1.0 (0.10)	18 1.1 (0.10)
OCT central (μm)	24 14.1 (2.7)	20 13.7 (2.1)	24 16.2 (4.5)	20 16.2 (4.4)
OCT Nasal inner (μm)	24 22.7 (2.7)	20 22.4 (2.4)	24 49.8 (4.1)	20 50.0 (4.4)
OCT Nasal outer (μm)	24 53.9 (10.1)	20 52.8 (8.5)	24 34.0 (3.7)	20 34.4 (3.7)
OCT Superior inner (μm)	24 26.9 (3.2)	20 26.6 (3.0)	24 50.3 (4.0)	20 48.8 (8.0)
OCT Superior outer (μm)	22 39.7 (5.6)	19 39.9 (5.3)	22 35.2 (4.0)	19 35.4 (4.2)
OCT Temporal inner (μm)	24 18.2 (1.7)	20 18.0 (1.9)	24 46.0 (4.9)	20 46.4 (4.3)
OCT Temporal outer (μm)	24 21.1 (1.6)	20 21.1 (1.6)	24 33.9 (4.0)	20 33.8 (3.5)
OCT Inferior outer (μm)	24 26.3 (3.0)	20 25.8 (2.7)	24 50.0 (4.5)	20 49.4 (4.1)
OCT Inferior inner (μm)	20 40.2 (7.5)	19 38.9 (7.3)	20 33.6 (3.1)	19 34.4 (4.1)

OCT, optical coherence tomography; RNFL, retinal nerve fiber layer; RGC, retinal ganglion cell layer.

NEI VFQ-25 and Macular Inner Retinal OCT Measures.

Univariate and multivariable linear regression (age and gender adjusted) revealed that VFQ-25 total was not associated (P > 0.05) with MoCA, CPC, or LBCRS. The VFQ-25 subscales and question #7 were also not associated with MoCA, CPC, or LBCRS (P > 0.05). Associations between macular inner retinal OCT measures and MoCA using univariate and multivariable analyses were not significant. However, the positive association between MoCA scores and right eye macular OCT RNFL volume, as noted in Figure 2, reached near statistical significance (P = 0.06). Associations between macular OCT measures and LBCRS were not significant using univariate analysis or multivariable analysis (P > 0.05), and none of the macular OCT measures revealed significant associations with the CPC by univariate or multivariable analyses (P > 0.05).

Figure 2.

Regression of macular OCT retinal nerve fiber layer volume for right eyes and Montreal Cognitive Assessment (MoCA) scores.

Abnormal MoCA versus normal MoCA.

Average VFQ-25 scores for those with normal MoCA scores (n = 18) was approximately one point higher than for those with abnormal scores (n = 10), but this was not statistically significant (95.5±3.7 versus 94.6±5.5, respectively, P > 0.05). Macular OCT RNFL volumes and inner and outer ring RNFL thicknesses were lower for each eye for those with abnormal MoCA scores compared to those with normal scores (Table 3), although not statistically significant. Except for left eye RGCL volumes and left eye outer ring RGCL thickness, macular RGCL volume and thicknesses for each eye were also lower in those with abnormal MoCA scores (Table 3).

Table 3.

Comparison of VFQ-25, OCT RNFL volume, and OCT inner and outer ring thickness measures for those with Montreal Cognitive Assessment (MoCA) scores less than 26 and greater than or equal to 26.

Measures	MoCA <26/30	MoCA ≥26/30	t-test p-value
VFQ-25
Patient totals	10	18
Mean (SD)	94.6 (5.5)	95.5 (3.7)	0.61
Median	96.0	96.9	0.85
Macular OCT volume mm3
Eye and Patient totals	RE 6 / LE 6	RE 14 / LE 12
Mean (SD)
Right eye RNFL	0.94 (0.14)	1.01 (0.12)	0.23
Left eye RNFL	0.99 (0.07)	0.99 (0.12)	0.49
Right eye RGCL	1.06 (0.13)	1.06 (0.09)	0.17
Left eye RGCL	1.06 (0.13)	1.06 (0.09)	1.00
Macular OCT inner ring thickness μm
Eye and Patient totals	RE 8 / LE 7	RE 16 / LE 13
Mean (SD)
Right eye RNFL	90.0 (8.3)	96.2 (8.5)	0.10
Left eye RNFL	88.6 (8.1)	95.1 (8.2)	0.11
Right eye RGCL	189.1 (17.9)	199.7 (15.3)	0.15
Left eye RGCL	190.9 (18.2)	196.7 (16.8)	0.48
Macular OCT outer ring thickness μm
Eye and Patient totals	RE 6 / LE 6	RE 14 / LE 12
Mean (SD)
Right eye RNFL	148.0 (23.7)	159.6 (21.4)	0.30
Left eye RNFL	150.0 (23.8)	155.7 (20.4)	0.61
Right eye RGCL	130.7 (11.5)	137.3 (14.1)	0.32
Left eye RGCL	139.7 (19.5)	137.2 (13.0)	0.76

RE, right eye; LE, left eye

Conclusions

In this exploratory study of participants from a healthy control group in the CO-AMD Registry, 36% of participants had abnormal cognitive testing in the range of mild cognitive impairment despite no known prior history of MCI or NDD and no cognitive complaints. These findings are consistent with investigations of the general population revealing 28-38% of people >65 years harbor unrecognized clinical manifestations of NDD^5,8,9. The withdrawal of one participant highlights the delicate nature of cognitive screening in healthy subjects. We adjusted our script to participants to emphasize that cognitive testing can be challenging, and this approach will be important for future studies.

NEI VFQ-25.

Drawings by four participants with abnormal cognitive testing demonstrated that significant visuospatial impairment can be revealed (Figure 1) in people without previously diagnosed MCI or NDD and with no cognitive complaints. Despite these findings, the VFQ-25, and its subscales for driving, near activities, peripheral vision, and question #7, were not associated with MoCA scores. Those with impairment on the MoCA had scores one point less on the VFQ-25 compared to those without cognitive impairment, which was not statistically significant. To detect a two-point difference on VFQ-25 composite scores between any two groups, 1568 participants per group is required¹⁸. Since these exploratory data revealed only a one-point difference VFQ-25 scores, the VFQ-25 is unlikely to be impacted by unrecognized cognitive impairment alone unless study group sizes are very large. Interestingly, VFQ-25 appears not to capture vision-specific changes in quality of life in participants with abnormal screening MoCA scores even when visuospatial dysfunction is present. A lack of insight (i.e., anosognosia) for decreased visual brain functioning could have contributed to these findings, particularly since participants had no cognitive complaints. LBCRS and CPC were also not associated with VFQ-25. Given that prior studies revealed decreased VFQ-25 in those with PD and other parkinsonian conditions¹², the lack of an association for those with parkinsonian findings could be due to the unexpected low prevalence of positive screening for parkinsonian signs and symptoms. In summary, further exploration of the impact of undetected cognitive impairment or NDD on VFQ-25 scores in age-related eye disease research appears to be unwarranted.

Macular OCT.

Those with cognitive impairment had macular inner retina OCT volumes and thickness measures 3-7% below those without impairment, for all but left eye RGCL volume and left eye RGCL outer ring thickness. Although differences were not significant between these small groups, we calculated a sample size for future studies. Assuming a power of 80%, an alpha of 5%, and prevalence of cognitive impairment of 33%, the estimated total sample size to detect a significance difference is 133 patients. If a future larger study supports these exploratory findings, then this range of differences in OCT inner retinal measures is consistent with the data available for studies of MCI cohorts versus controls¹⁹. Once again, when determining whether screening for NDD or cognitive impairment is necessary for age-related eye disease research, sample size and group characteristics (e.g., age and disease-associations with NDD) must be considered. Given the growing data revealing retinal thinning in a variety of age-related NDD and dementing illnesses, we believe these results suggest that further investigation is warranted to understand whether MoCA screening will be important to control for undiagnosed cognitive impairment on retinal OCT measures within age-related eye disease cohorts.

A major limitation of our exploratory study is the small sample size. However, the prevalence of abnormal cognitive screening in age-related eye disease research cohorts was not previously investigated prior to this study, and visual outcome measures for those screened for NDD or cognitive impairment were also not previously investigated. For these reasons, the study provides data that allowed calculation of sample sizes for future studies related to OCT measures and determine that VFQ-25 differences are unlikely to yield important differences in future studies with groups <1600. Another limitation is the lack of inclusion of an eye disease study group (i.e., those with AMD), which is particularly important given the higher rate of dementia associated with AMD²⁰. Future studies should incorporate the use of a cognitive screening measures for the visually impaired, and inclusion of blood-based biomarkers such as beta-amyloid and tau are worth considering in future studies if sensitivity and reliability improve.

In summary, further investigations with a larger cohort using sample sizes from the estimates noted and inclusion of those with age-related eye disease will be necessary to confirm the potential associations between cognitive testing and macular OCT measures, but further investigations do not appear warranted for the VFQ-25 outcome measure. Ultimately, future studies could discern whether it is necessary to screen for unrecognized cognitive impairment as a confounder in age-related eye disease research.