Altered Vergence Eye Movements and Pupil Response of Patients with Alzheimer’s Disease and Mild Cognitive Impairment During an Oddball Task

Abstract

Background:

Alzheimer’s disease (AD) is characterized by progressive deterioration of cognitive functions and may be preceded by mild cognitive impairment (MCI). Evidence shows changes in pupil and vergence responses related to cognitive processing of visual information.

Objective:

Here we test the hypothesis that MCI and AD are associated with specific patterns in vergence and pupil responses.

Methods:

We employed a visual oddball task. In the distractor condition (80%of the trials), a blue stimulus was presented whereas in the target condition (20%of trials) it was red. Participants (23 Controls, 33 MCI patients, and 18 AD patients) were instructed to press a button when a target appeared.

Results:

Participants briefly converged their eyes 200 ms after stimulus presentation. In controls, this transient peak response was followed by a delay response to targets but not to distractor stimuli. In the patient groups, delay responses to distractors were noticed. Consequently, the differential vergence response was strong in the control group, weak in the MCI group, and absent in the AD group. Pupils started to dilate 500–600 ms after the appearance of a target but slightly contracted after the presentation of a distractor. This differential pupil response was strongest in the AD group.

Conclusion:

Our findings support the idea of a role of vergence and pupil responses in attention and reveal altered responses in MCI and AD patients. Further studies should assess the value of vergence and pupil measurements as an objective support tool for early diagnosis of AD.

Keywords

Attention binocular eye movement convergence dementia Edinger-Westphal fixational eye movement locus coeruleus mild cognitive impairment neurodegeneration oculomotor behavior

INTRODUCTION

Alzheimer’s disease (AD) it is a neurodegenerative illness, manifested by a progressive cognitive impairment and behavioral disorders. AD occurs most frequently (but not exclusively) in people over 65 years of age, and people with mild cognitive impairment (MCI) are at an increased risk of developing AD. Currently, there are 50 million people in the world with an AD diagnosis and it is expected to rise to more than 115 million people world-wide by 2050 [1]. Although the precise etiology of AD is not yet clearly understood, AD results in severe synaptic and neuronal loss [2]. It has been suggested that amyloid-β (Aβ) deposition, presence of tau protein, and production of neurofibrillary tangles are responsible for the synaptic and cortical atrophy [3].

The neurodegenerative process not only causes loss of episodic memory but other cognitive functions are affected as well, such as language, executive functions, attention, and visuospatial functioning [4 –7]. For instance, AD patients have reduced ability to orient attention toward relevant visual input, possibly as a result of a weakened dorsal attention network [8]. AD patients also show visual and ocular movement problems [9]. Visual dysfunctions include decreased visual acuity, poor color discrimination, and visual field loss [10]. Oculomotor deficits are broad consisting mainly of different saccade metrics compared to healthy controls [10, 11], altered pupil responses [12], and smaller and irregular eye vergence movements [13].

Recently, studies showed that the locus coeruleus is the first region to develop pathology precursors and may be a critical step in the pathogenesis AD [14, 15]. The locus coeruleus is a small nucleus located in the pons of the brainstem and forms part of the reticular activating system. It is the main site for the synthesis of norepinephrine, and via its widespread connections throughout the brain the locus coeruleus influences many brain functions, such as arousal, blood flow, heart rate, emotions, and sleep-wake cycle [16]. As the main source of norepinephrine to the pre-frontal cortex, the locus coeruleus has also an impact on cognitive processes that sub-serve executive functions [16 –21]. Besides cognitive areas, the locus coeruleus modulates activity of motor-related neurons in the rostral pole of the superior colliculus [22, 23], which have a role in horizontal disconjugate eye movements or vergence [24]. The noradrenergic influence in these regions depends on the activity of the locus coeruleus. Neurons in the locus coeruleus fire in two distinct modes: tonically or with short phasic bursts. Tonic firing is implicated in the regulation of behavioral state and state-dependent cognitive processes [25] whereas task relevant stimuli induce a phasic response [26] implicated in the P300 event-related potential [27, 28]. Among others, input to the locus coeruleus comes from structures involved in horizontal eye movement control, such as the medial prefrontal cortex [29, 30] and the nucleus prepositus [31].

Within the brainstem, the locus coeruleus connects to the Edinger-Westphal nucleus [32]. The Edinger-Westphal nucleus comprises of two distinct populations: the preganglionic oculomotor neurons and the centrally projecting neurons. Pre-ganglionic neurons project to postganglionic neurons in the ciliary ganglion, which in turn innervate smooth muscle fibers in the sphincter muscle of the iris to control pupil size. The centrally projecting Edinger-Westphal cells are involved in non-ocular functions such as feeding behavior and stress responses, and project to the brainstem, the spinal cord, and prosencephalic regions. The Edinger-Westphal preganglionic neurons receive input from the olivary pretectal area conveying pupillary light reflex input (see [32]), and from neurons of the central mesencephalic reticular formation [33] that have a role in vergence eye movements [34]. Also, the superior colliculus is reciprocally connected to the Edinger-Westphal [32, 35]. The Edinger-Westphal cells further receive monosynaptic excitatory input from ventral hippocampal cells, which are critical for attention [36]. These hippocampal cells innervate the medial prefrontal cortex. This area is known to control attention processing [36] and eye vergence (e.g., [30]).

Recent evidence shows a relation between eye vergence movements and cognitive information processing. In a series of studies, we have demonstrated that subjects briefly converge their eyes when orienting attention to a visual or auditory stimulus [37 –40]. These vergence responses predict whether or not a visual stimulus will be perceived and be held in short-term memory [41 –45]. The strength of these attention-related eye vergence responses positively correlates with the strength of late components of visual event related potentials in parietal areas [40]. Eye vergence may cause changes in pupil size [46] that manifest the modulation of cognitive processing of visual information, including mental effort, attention, memory, and perception [47, 48]. Interestingly, pupil fluctuations track changes in locus coeruleus activity [20, 49] and patients with AD have abnormal pupillary function compared with such function in healthy aging individuals [50]. Thus there is a link relating vergence eye movement and pupil responses with cognitive processing.

The aim of the current study is to test these eye responses in MCI and AD patients, and to evaluate whether they correlate with cognitive decline. We therefore examined the vergence and pupil responses of patients diagnosed with AD and a group of patients with MCI while performing an oddball task and compared their recorded vergence and pupil data to those of cognitive healthy controls. Previous studies have shown that an oddball paradigm induces pupil responses, most likely mediated by locus coeruleus activity [20, 51].

The results show that after stimulus presentation, subjects briefly converge their eyes. In the subjects of the control group, this transient vergence response was followed by a delay response to targets but not to distractor stimuli. Patients, particularly of the AD group, showed a delay response to targets as well as to distractors. Consequently, the differential vergence response was strong in the control group, weak in the MCI group, and absent in the AD group. Pupils dilated after the appearance of a target but contracted after the presentation of a distractor. This differential pupil response was strongest in the AD group. Our data confirms earlier reports on altered pupil responses in patients suffering from dementia [12] and agree with findings of eye vergence and pupil responses during attention processing [37 –40]. The gradual change in vergence and pupil modulation from cognitive healthy controls to MCI to AD may indicate that these eye metrics have the potential to be valuable as an objective marker for early AD detection.

MATERIALS AND METHODS

Participants

We tested 74 subjects (34 men and 40 women). Some of them were diagnoses with MCI (N = 33) or with possible AD (N = 18). The rest of the subjects (N = 23) were cognitively healthy controls. The Mini-Mental State Examination (MMSE) scores ranged between 28 and 30 for the healthy controls and 23–30 for the MCI patients, and 14–26 for AD patients, out of a possible 30 points. The Montreal Cognitive Assessment (MoCA) scores ranged between 25 and 30 for the healthy controls and 18–28 for the MCI patients, and 16–22 for AD patients, out of a possible 30 points. The general demographic and clinical features of the groups are shown in Table 1.

Table 1

Table1

	Control (n = 23) Mean (SD)	MCI (n = 33) Mean (SD)	AD (n = 18) Mean (SD)	Control versus MCI	Control versus AD	MCI versus AD
Age (y)	67.1 (8.6)	70.3 (6.8)	74.1 (8.3)	0.14	0.053	0.031
Gender (M/F)	6/17	15/18	5/13
GDS	1.19 (0.5)	2.91 (0.52)	3.92 (0.51)	< 0.00001	< 0.00001	< 0.00001
MoCA	27.8	23.4	19,0	0.0008	0.00003	0.0017
MMSE	29.3	26.6	23,8	0.056	0.0005	0.036

GDS, Global Deterioration Scale; MoCA, Montreal Cognitive Assessment; MMSE, Mini-Mental State Examination. For the comparison of these variables, we used Kruskal-Wallis test. Post-hoc contrasts were significant at p < 0.016 (Bonferroni-corrected).

The exclusion criteria were as follows: 1) Severe cognitive deterioration, operationalized as MMSE below 10 [12, 52]; 2) history of neurological disease with clinically relevant impact on cognition (e.g., cerebrovascular disease); 3) severe psychiatric disorder; 4) incidental structural brain findings with impact on cognitive impairment (e.g., brain tumor); 5) presence of relevant visual problems; and 6) problems for understanding spoken or written Spanish language.

Ethics statement

Participants and their families received detailed instructions for the experiments. Before entering the study, patients or relatives signed a written informed consent for their participation in accordance with the Helsinki Declaration. The Ethics Committees of the University of Barcelona and Hospital Sanitas CIMA approved the study.

Clinical assessment

All patients were evaluated by experienced neurologists, neuropsychologists and radiologists (all patients received a multi-sequence 1.5 T MRI assessment). The diagnosis of ‘control’, ‘MCI’, or ‘AD’ was established at consensus meetings with neurologists and neuropsychologists, following the recommendations of the National Institute on Aging-Alzheimer’s Association [53]. A ‘core clinical criteria’ for the MCI was made when the following criteria were fulfilled: 1) concern regarding a change in cognition, from the patient, a proxy informant or a skilled clinician; 2) impairment in one or more cognitive domain is shown (performance is typically below 1–1.5 SD), according to participant age and education; 3) preservation of independence in functional abilities; and 4) not demented. When the decline in one or more cognitive domains exceeded these criteria and there was a decline in functional activities, with exclusion of other potential causes of dementia, patients were diagnosed as possible AD.

Neuropsychological testing

Participants underwent a comprehensive neuropsychological battery including the following tests: Cognitive performance: MMSE, Free and Cued Selective Reminding Test, Rey–Osterrieth Complex Figure, and phonological and semantic verbal fluency; Depression and Anxiety: Geriatric Depression Scale, State-Trait Anxiety Inventory; Functional scales: Clinical Dementia Rating Scale, Functional Activities Questionnaire. Age- and education- adjusted values for the different neuropsychological variables were obtained from the Spanish Normative Studies (NEURONORMA) Project [54, 55].

Device

We used the BGaze (Braingaze SL, Mataró, Spain) system to present the visual stimuli and record eye position and pupil data. Screen resolution was 1024× 768 pixels, and the employed remote eye tracker was an X2-30 (30 Hz, Tobii Technology AB, Sweden). According to the manufacturer, the accuracy under ideal conditions for binocular viewing is 0.4° of visual angle, and the precision 0.32° of visual angle. Note that random variability of noise is not a serious problem since it will be averaged close to zero over a series of trials [56].

Procedure

The task was carried out in a room of the hospital with the lights turned off and the curtains closed, in order to have dim lighting conditions. The subjects were seated approximately 50 cm from the screen where stimuli were presented and the eye-tracker was placed below the screen. Patients could wear corrective lenses. A chinrest was used to avoid large head movements.

Paradigm

An oddball paradigm was applied. An oddball task activates many brain regions (see https://neurosynth.org/analyses/terms/oddball/) and is used to assess neural activity in the locus coeruleus [19]. The task was similar to the one described in [57] to obtain an ERP P300 component in MCI and AD patients. Also, we opted for a visual oddball paradigm as we are interested in oculomotor measures. The experimental task consists of a sequence of 100 trials. In each trial, a grey screen (Mask) was presented for 2000 ms. Then a visual stimulus was centrally presented for 2000 ms (Fig. 1). The central stimulus was a series of letters forming a string of 11 characters. Characters were randomly selected and could be in upper or lower case. They did not represent acronyms or meaningful words in order to avoid possible bias. Except for color, the strings were comparable. In one of the experimental conditions, all characters were in blue font (80%of trials), whereas in the other condition they were in red font (20%of trials). The participants were instructed to look at the screen and to press a button only when the characters appeared in red color. Instructions were repeated verbally by the experimenter during the task if necessary. Character strings in red were considered targets while strings in blue font represented distractors. Stimuli were randomly presented. During the task, eye movements were recorded by a remote eye tracker. The eye tracking equipment was calibrated (5 points, binocular) for each participant at the beginning of the experiment. The total duration of the task was about 6 min.

Fig. 1

Oddball task scheme. Strings of symbols are presented for two seconds, followed by a grey mask. In 80%of the trials the symbols were printed in blue, and in the remaining 20%they were printed in red (oddball).

Vergence and pupil response analysis

The eye data obtained during the oddball task was used to calculate vergence eye movements and pupil responses. Before calculating these eye metrics, preprocessing of the positional eye data was done. Recorded data points which did not correspond to valid pupil detections (i.e., whenever the validity score given by the eye-tracker software had a non-zero value) were marked out. Trials containing too many invalid data points (20 points or more) were discarded. The exclusion rates were 38%, 28%, and 53%for Control, MCI, and AD, respectively. The corresponding standard deviations were 43%, 31%, and 37%. T-tests show that the difference MCI-AD is significant (p = 0.02) but the others not (p > 0.28). A significant difference in the number of included trials may have altered the signal-to-noise ratio confounding the findings. To assess within-subject variability, we evaluated the RMS-means (c) of the standard deviations of the averaged vergence response in our ‘delay-response’ window (see below) per subject, separately calculated for trials with distractors and for trials with targets. For the averaged vergence responses c = 0.40 for distractors and c = 0.39 for targets. The same but separated into groups c = 0.40, 0.40, and 0.31, (distractors) and c = 0.36, 0.40, and 0.34 (targets) for Control, MCI, and AD, respectively. Eventually, interpolation was applied in order to produce sequences of evenly spaced points. Afterwards, Gaussian smoothing took place (about general guidelines on preprocessing, see, e.g., [78]). In order to calculate vergence changes, we transformed the coordinates of left and right eye, supplied by the eye tracker, into angular magnitudes. Rather than the angle of vergence itself —say γ—, we are focusing on the relative vergence modulation $υ (t) \equiv \frac{γ (t) - γ_{o}}{max | γ (t) - γ_{0} |}$ , where γ₀ stands for the γ value at stimulus onset time, and the indicated maximum was taken for all the absolute values of the difference γ(t)-γ₀ in every considered trial. The subtraction of initial values from every response, which was applied both to vergence and to pupil data, served the purpose of obtaining relative changes. Afterwards, all the υ (t) sequences coming from trials in the same condition (target, distractor) and group (Control, MCI, AD) were averaged in order to obtain ‘mean υ (t)’ curves. Only correct trials (i.e., when participants correctly responded to targets or refrained from responding to distractors) were analyzed. To estimate a correlation between pupil and vergence responses, we calculated the averaged cross-correlations. For the statistical analysis, we performed a number of comparisons based on the t-test, and an Analysis of Variance (ANOVA) from all the accepted trials. We tested the trial distributions for normality. The data showed that not all distributions were normal. However, permutation tests produced similar outcomes as t-tests. As post-hoc test, we applied the Tukey-Kramer procedure.

RESULTS

Behavioral responses

We first looked at the behavioral performance in the oddball task. Using signal detection theory language, a detection of a target can be called ‘hit’, while the absence of response in the presence of a distractor is viewed as a ‘correct rejection’. Adding the two associated errors, we obtain hits, correct rejections, misses, and false positives. The corresponding rates for these four scores were (0.86, 0.99, 0.14, and 0.01) for the Control group, (0.70, 1.0, 0.30, and 0.00) for the MCI group, and (0.52, 0.94, 0.48, 0.06) for the AD group. Hit rates involve significant differences (F_2,71 = 3.3 p = 0.04, but post-hoc indicates that only the Control-AD difference was significant). Thus, behavioral performance is lowest for the group with most severe impairment.

Vergence

We next analyzed the vergence responses to distractors (i.e., the blue stimuli) and to targets (i.e., the red stimuli). Figure 2 (top) gives an example of the average responses to distractors and targets for a control subject, an MCI patient, and an AD patient. The curves show an initial peak response (clearly visible in the responses to distractors of the control group). After the initial peak, responses start to slowly decline and sometimes stay relatively stable during the rest of the trial, especially in the distractor trials. We refer to these responses as delay responses. The averaged vergence responses from all participants are shown in Fig. 2 (bottom; sample-wise tests were not corrected for multiple comparisons). In all groups and for both conditions, responses start around 200 ms after stimulus onset. To observe differences in response latencies, we calculated the peak latencies of the average vergence responses. Vergence responses in the control group have peak latencies of 533 ms (distractors) and 767 ms (targets). In the MCI and AD groups, the latencies were somewhat longer (MCI: 567 ms for distractors and 867 ms for targets; AD: 700 ms for both distractors and targets).

Fig. 2

Top: Examples of the subject-averaged vergence curves. Left panel shows a Control subject, and the numbers of included trials were 78 distractor trials and 20 target trials. Mid panel shows am MCI patient with 55 distractor trials and 14 target trials. On the right, an AD patient, with 67 distractor trials and 16 target trials is shown. Shaded areas correspond to the 95%CI bound around the curve. Time is from stimulus onset. Bottom: The grand-averaged curves for the relative vergence modulation in the three subject groups (Control, MCI patients, and AD patients) during the oddball experiment. The grey bars indicate the domains where the difference between distractor (blue) and target (red) mean curves is significant (paired t-test, for a limit of α= 0.05). Shaded areas correspond to the 95%CI bound around the curve. Time is from stimulus onset.

Fig. 5

Strength of the delay pupil response calculated over the window 1000 ms –2000 ms after stimulus presentation. Error bars are SEM. Note the horizontal lines within the boxes corresponding to the median values.

The delay responses to distractor stimuli are very weak in the control group but clearly present in the two patient groups. Though, the delay response is stronger in the group of AD patients than in the MCI group. Thus, stimulus evoked vergence responses following the initial peak response are somewhat higher in the AD group than in the MCI group. Because the noticeable differences occur during the delay period, we computed the averaged responses for the window 1000–2000 ms after stimulus presentation (Fig. 3). More elaborate ways of selecting windows of interest have been discussed by Brooks et al. 2017 [79]. In the control group, the average responses to distractors (–0.01±0.51) was significantly (p < 0.001) weaker than the responses to targets (0.24±0.50). This was also true for the MCI group (distractors: 0.09±0.46, targets 0.28±0.45, p < 0.001) but not for the AD group (distractors 0.17±0.44, targets 0.27±0.51, p = 0.09). The stimulus type (distractor/target) is a significant factor (F_1,3827 = 53.43, p < 0.001). The subject group (Control/MCI/AD) is also significant (F_2,3827 = 6.18, p≈0.002). However, the interaction between both factors approached but did not reach statistical significance (F_2,3827 = 2.57, p = 0.08). Considering subject averages of vergence responses for distractor stimuli and target stimuli, the stimulus type factor yields means of 0.06±0.25 and 0.25±0.27 for distractors and targets respectively (F 1,105 = 16.36, p≈10^–4).

Fig. 3

Strength of the delay vergence response calculated over the window 1000 ms –2000 ms after stimulus presentation. Error bars are SEM. Note the horizontal lines within the boxes corresponding to the median values.

To further explore the interaction, we computed a modulation index. The modulation index (m) was obtained by calculating the difference between the responses to distractors and targets over the time interval of 1000–2000 ms. Given a pair of mean curves for targets and distracters, say V _T (t) and V_D(t), m is the average of the quantity ${[\frac{υ_{T} (t) - υ_{D} (t)}{υ_{T} (t) + υ_{D} (t)}]}^{+}$ for all the t values in the considered window, with [x]⁺ ≡ x for x≥0 and 0 otherwise. We found the averages m = 1.19, 0.52 and 0.21 for Control, MCI and AD groups, respectively. Thus, according to this descriptive analysis, the mean modulation curves for hits (targets) and for correct rejections (distractors) get closer to each other when going from cognitive healthy to MCI and AD groups (see Fig. 2).

Pupil responses

We calculated the relative modulation of the pupil size from the pupil size measurements given by the eye tracker device. The obtained pupil response curves are shown in Fig. 4. Pupil responses start around 500–600 ms after stimulus presentation. The behavior of the responses is different to targets than to distractors. To targets, pupil size increases (dilates) in particular in AD patients whereas pupils become smaller after presenting distractor stimuli.

Fig. 4

Relative modulation of pupil response to distractors and targets. Note the presence of larger relative increases for AD patients. The grey bars indicate the domains where the difference between blue and red mean curves is significant. Shaded areas correspond to the 95%CI bound around the curve. Time is from stimulus onset.

We calculated the peak latencies of the pupil responses (average of both eyes) to targets. For the control group, peak latency was 1200 ms, for the MCI group it was 1416 ms, and for AD patients 1950 ms. The latencies to minima for the case of distractors were 1600 ms (control group), 1867 ms (MCI group), and 1883 ms (AD group).

We then assessed the strength of the delayed pupil responses by averaging the mean pupil size curves for the window 1000–2000 ms. In all groups, the average pupil responses to distractors were significantly (p < 0.001) weaker than the responses to targets (Fig. 4; control group: distractors, –0.07±0.43, targets, 0.18±0.43; MCI group: distractors –0.11±0.43, targets: 0.19±0.44; AD group: distractors, 0.19±0.40, targets, 0.36±0.36). The stimulus type (distractor/target) is a significant factor (F_1,3827 = 147.53, p < 0.001). The subject group (Control/MCI/AD) is not significant (F_2,3827 = 0.62, p = 0.54) but the interaction between both factors is (F_2,3827 = 6.31, p = 0.002). Group comparisons for fixed stimulus type indicate significance in the differences AD-MCI and AD-Control, which is higher for distractor stimuli.

Considering subject averages of pupil responses for distractor stimuli and target stimuli, the stimulus type factor yields means of –0.10±0.18 and 0.16±0.20 (F 1,105 = 53.99, p = 4.5·10^–11). In addition, the category factor leads also to significant differences (F 2,105 = 3.75, p≈0.03). Post-hoc indicates that this significance comes from the MCI-AD difference [means –0.03±0.23 (MCI) and 0.05±0.29 (AD)].

Task duration

To evaluate a possible effect of task duration, we compared the average maximum values of the peak responses and average delay responses of the trials from the first one-third of task with those of the last one-third of the task. Significant differences between the maxima of the vergence responses of the MCI (F₇₃₇ = 5.83, p < 0.01) and AD (F₂₁₂ = 5.28, p < 0.02) groups were observed. All other vergence and pupil comparisons yield no significant results.

Correlation between responses and disease severity

We calculated the correlations (Spearman’s ρ (’rho’)) between maximum and mean delay vergence/pupil responses and the outcomes of the screening tests (GDS, MoCA, and MMSE). Significant correlations are found between maximum and delay vergence responses to targets and MoCA scores (mean: ρ= 0.55; p = 0.02; maximum: ρ= 0.63; p = 0.01), between pupil responses to distractors and MMSE (ρ= 0.72; p = 0.02), and pupil responses to targets and GDS (ρ= 0.30; p = 0.04). All the other correlation analyses (see Supplementary Material) gave no significant results.

Correlation between vergence and pupil responses

Even though the observed response patterns of vergence and pupil size are very different, a relation between both components has been described [46]. To assess such a correlation, we calculated the cross-correlation between mean vergence and pupil curves (averaged for left eye and right eye), which can be viewed as a function of a time lag (τ). The results show that for distractors there is anti-correlation, which is strongest in the AD group (Fig. 6). For targets, there is a positive correlation between vergence and pupil responses, which again is strong in AD group (Fig. 6).The peaks of the correlation curves have all a positive time lag indicating that vergence responses precede pupil responses by about 600–1000 ms. This agrees with the idea that vergence is causal to pupil response [46]. As purely descriptive figures, the mean lag times for these peaks are 567 ms, 733 ms, and 967 ms for Control, MCI, and AD, respectively.

Fig. 6

Cross-correlations between the mean curves for vergence and pupil responses.

Definiteness in the delays can be measured by the peak widths. We have evaluated the broadness of these curves by finding their width at a height of one third of the peak value. Considering Fig. 6, in the control and MCI groups these widths amount to 1967 ms. The correlation curve of the AD group has a peak width of 2333 ms. The broader peak in the AD group may indicate a less specific correlation.

DISCUSSION

We assessed eye vergence and pupil responses during a visual oddball paradigm to evaluate these items in cognitive healthy subjects, and in MCI and AD patients. The oddball paradigm is commonly used to study effects of stimulus novelty and attention. Presentation of a visual stimulus results in a vergence and pupil response after 200 ms and 500–600 ms, respectively. The patient groups showed longest delays in peak latencies of vergence and pupil responses. So vergence responses clearly precede the pupil responses, which may be explained by the idea that a vergence movement triggers a pupillary response [46].

Vergence and pupil responses may be affected by the arousal of the subject, which partly could explain the differences between control and patient groups. In line with this idea, we obtained differences in maximum vergence response during the task in the MCI and AD groups and not in the control group. We did not find alterations in pupil responses during the task, indicating that the functioning of the locus coeruleus, which is critical in arousal and can be indexed by pupil size [49, 60] was not altered during the task. This would indicate that cognitive arousal of the subject is unlikely to have a major impact on the findings.

The behavior of the responses depends on the stimulus condition (target or distractor) and patient type (control, MCI, or AD). In the control group, distractors only elicit a transient vergence response whereas targets produce a sustained vergence response as well. MCI and particularly AD patients also show delay vergence responses to distractor stimuli. As a consequence, the difference between the delay responses to distractors and targets, which is evident in the control group, is weaker in the MCI group and absent in the AD group. Pupils dilate when the stimulus is a target but slightly constrict when the stimulus is a distractor. This differential response happened in all three groups but was more pronounced in the AD group. The finding of pupil dilation in target trials may be related to task demand, which leads to increases in pupil size [48] that are associated with poor performance [58]. However, increases in pupil response can also be due to novelty effect [59].

Pupil size decreases significantly with age, in particular in AD [61]. This may be an effect of the degenerating of the locus coeruleus, which is known to drive pupillary response by its noradrenergic input to the Edinger-Westphal nucleus. In our study, we observed strongest pupil dilation in target trials in AD patients, which is at odds with a less active locus coeruleus. An explanation for the differential responses is that the Edinger-Westphal nucleus is part of the neural circuitry of attention. For instance, inactivation of neurons in the Edinger-Westphal nucleus impairs behavioral performance in a visual-detection task that demands attention [36]. Pre-ganglion neurons of the Edinger-Westphal nucleus receive direct input from the ventral hippocampus [36], which connects to the prefrontal cortex. Prefrontal cortex enhances activity to relevant items and inhibits irrelevant activity. Such attentive behavior we see in the enhanced vergence and pupil responses to targets and suppressed responses to distractors. Prefrontal cortex also has a role in controlling vergence eye movements [30, 62], and activation of prefrontal neurons causes dilation of the pupils [36, 64]. Moreover, the Edinger-Westphal is reciprocally connected to the superior colliculus [32, 35], which integrates visual, auditory and somatosensory information to initiate eye movements. Thus, the Edinger-Westphal nucleus may integrate attention signals with vergence and pupil commands producing the observed differential responses. This idea also could offer an explanation for the finding that an auditory cue results in a vergence response [37] and that pupils dilate in auditory oddball tasks (e.g., [65]). Whether an auditory oddball task produces a vergence response needs to be tested. However, AD patients typically show frontal-hippocampal atrophy indicating that pupil dilation should be reduced and not enhanced as we observe in our study. Besides sympathetic activation, pupil dilation can be the results from para-sympathetic inhibition [66]. There is a direct projection from the hypothalamus to the Edinger-Westphal nucleus suppressing pupil dilation. Dysfunctions of the hypothalamus in AD patients may therefore mediate pupil dilation.

An important covariate, which significantly influences the pupil responses is medication acting either on adrenergic or on acetylcholine receptors. In our study, all AD patients were treated with acetylcholinesterase inhibitors. Controls and MCI patients did not receive medication that could affect the autonomous nervous system (both sympathetic and parasympathetic). Therefore, medication may affect pupil responses in the AD group.

Neurophysiological considerations

The P300 event-related potential (ERP) represents cognitive and attentive processing of sensory information. ERP studies show that an oddball task elicits P300 responses across multiple cortical, thalamic, and limbic regions [67, 68]. The P300 amplitude varies with the probability of the target and its onset latency with the difficulty of the task (see [28]). In an oddball paradigm, AD and MCI patients manifest a delay of the latency [57 , 70] and a reduced peak [71, 72] of the P300 response. Our observation of longer peak latencies for the vergence and pupil responses found in MCI and AD patients are in line with the reported P300 observations in these patient groups. Also, our data agree with the reported existence of a positive correlation between the latency and amplitude of the late ERP components with latency and strength of the vergence responses in an attention task [40]. Noradrenergic neurons of the locus coeruleus are activated by target stimuli in an oddball paradigm [17 –19]. The P300 event-related potential is produced by phasic (and not tonic) activity of the locus coeruleus [27, 28], which enhances cortical encoding of salient stimuli [27]. The phasic response induces gamma power increase in the prefrontal cortex and is essential for mediating sensory information [26]. In this study, we describe a correspondence between vergence and pupil responses and attentive processing in an oddball paradigm. Considering the close relation between vergence and pupil responses with the different components that constitute or represent the attention network, we propose that the assessment of vergence and pupil responses in combination with an attention paradigm may reveal the function of the underlying attention network.

Marker for early AD diagnosis

Early diagnosis and monitoring of disease progression have become vital in clinical practice for optimal patient management. The current diagnosis of AD is based on the history and clinical observation of the patient taking into account the neurological and neuropsychological characteristics. Current biomarker measures are costly and invasive and cheaper non-invasive solutions applicable in routine clinical practice are desired.

The eye and the brain have an embryological similar origin. In fact, the retina is part of the central nervous system and can be considered as brain tissue. Also, Aβ plaques are present in the retina of AD patients, which even precede the presence Aβ plaques in the brain [73]. Ophthalmologic tools may therefore offer an alternative approach for evaluating the neurodegenerative process in clinical practice. For example, retinal imaging may potentially be a useful tool to detect AD [74]. Also, tracking eye gaze behavior supposes a potential biological marker of cognitive impairment [75]. Here we present evidence linking altered pupil and vergence responses to cognitive impairment in MCI and AD. The findings point to a gradual change in these metrics as the neurodegenerative process progresses. Thus vergence/pupil assessment may reveal the severity of cognitive dysfunction in these patients.

Tau lesions in the locus coeruleus, which may be the first identifiable pathology of AD [14, 15], negatively impact cognition [76, 77]. In view of the close association between pupil and vergence responses during a cognitive task with the functioning of the locus coeruleus, then the assessment of these items could be a potential candidate to consider as an objective non-invasive biomarker tool for the early diagnosis of AD. In a follow-up study, we apply machine learning models to detect differences in vergence and pupil responses at single-subject level (Hashemi et al., submitted). A critical next step will be then to test association of these markers with known AD biomarkers (e.g., Aβ, tau) and demonstration of normal vergence/pupil responses during a cognitive task in patients with non-AD dementias of comparable overall severity.

Footnotes

ACKNOWLEDGMENTS

This research was supported by grants from Spanish Ministry of Science (PGC2018-096074-B-I00) to HS, and from the University of Guadalajara to ECJ.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Prince

, Wimo

, Guerchet

, Ali

G-C

, Wu

Y-T

, Prina

(2015) The World Alzheimer Report 2015, The Global Impact of Dementia: An analysis of prevalence, incidence, cost and trends. Alzheimer’s Disease International (ADI), London.

Morrison

, Baxter

(2012) The ageing cortical synapse: Hallmarks and implications for cognitive decline. Nat Rev Neurosci 13, 240–250.

Sheppard

, Coleman

(2020) Alzheimer’s Disease: Etiology, Neuropathology and Pathogenesis. In Alzheimer’s Disease: Drug Discovery, Huang X, ed. Exon Publications, Chapter 1, Brisbane.

Scinto

, Daffner

, Castro

, Weintraub

, Vavrik

, Mesulam

(1994) Impairment of spatially directed attention in patients with probable Alzheimer’s disease as measured by eye movements. Arch Neurol 51, 682–688.

Perry

, Hodges

(1999) Attention and executive deficits in Alzheimer’s disease. A critical review. Brain 12, 383–404.

Weintraub

, Wicklund

, Salmon

(2012) The neuropsychological profile of Alzheimer disease. Cold Spring Harb Perspect Med 2, a006171.

Malhotra

(2018) Impairments of attention in Alzheimer’s disease. Curr Opin Psychol 29, 41–48.

, Wu

, Fleisher

, Reiman

, Chen

, Yao

(2012) Attention-related networks in Alzheimer’s disease: A resting functional MRI study . . Hum Brain Mapp 33, 1076–1088.

Garbutt

, Matlin

, Hellmuth

, Schen

, Johnson

, Rosen

, Dean

, Kramer

, Neuhaus

, Miller

, Lisberger

, Boxer

(2008) Oculomotor function in frontotemporal lobar degeneration, related disorders and Alzheimer’s disease. Brain 131, 1268–1281.

10.

Javaid

, Brenton

, Guo

, Cordeiro

(2016) Visual and ocular manifestations of Alzheimer’s Disease and their use as biomarkers for diagnosis and progression. Front Neurol 7, 55.

11.

Kapoula

, Yang

, Otero-Millan

, Xiao

, Macknik

, Lang

, Verny

, Martinez-Conde

(2014) Distinctive features of microsaccades in Alzheimer’s disease and in mild cognitive impairment . Age 36, 535–543.

12.

Granholm

, Panizzon

, Elman

, Jak

, Hauger

, Bondi

, Lyons

, Franz

, Kremen

(2017) Pupillary responses as a biomarker of early risk for Alzheimer’s disease. J Alzheimers Dis 56, 1419–1428.

13.

Uomori

, Murakami

, Yamada

, Fujii

, Yoshimatsu

, Nakano

(1993) Analysis of gaze shift in-depth in Alzheimer‘s Disease patients. IEICE Trans Inf Syst E76D, 963–973.

14.

Mravec

, Lejavova

, Cubinkova

(2014) Locus (coeruleus) minorisresistentiae in pathogenesis of Alzheimer’s disease. Res 11, 992–1001.

15.

Betts

, Cardenas-Blanco

, Kanowski

, Spottke

, Teipel

, Kilimann

, Düzel

(2019) Locus coeruleus MRI contrast is reduced in Alzheimer’s disease dementia and correlates with CSF Aβ levels. Alzheimers Dement 11, 281–285.

16.

Poe

, Foote

, Eschenko

, Johansen

, Bouret

, Aston-Jones

, Harley

, Manahan-Vaughan

, Weinshenker

, Valentino

, Berridge

, Chandler

, Waterhouse

, Sara

(2020) Locus coeruleus: A new look at the blue spot. Nat Rev Neurosci 21, 644–659.

17.

Foote

, Berridge

, Adams

, Pineda

(1991) Electrophysiological evidence for the involvement of the locus coeruleus in alerting, orienting and attending. Prog Brain Res 88, 521–532.

18.

Rajkowski

, Kubiak

, Aston-Jones

(1994) Locus coeruleus activity in monkey: Phasic and tonic changes are associated with altered vigilance. Brain Res Bull 35, 607–616.

19.

Krebs

, Park

HRP

, Bombeke

, Boehler

(2018) Modulation of locus coeruleus activity by novel oddball stimuli. Brain Imaging Behav 12, 577–584.

20.

Murphy

PR.

, O’Connell

, O’Sullivan

, Robertson

, Balsters

(2014) Pupil diameter covaries with BOLD activity in human locus coeruleus. Hum Brain Mapp 35, 4140–4154.

21.

Aston-Jones

, Rajkowski

, Kubiak

, Alexinsky

(1994) Locus coeruleusneuronsin monkey are selectively activated by attended cues in a vigilance task. J Neurosci 14, 4467–4480.

22.

Kayama

, Sat

(1982) Effects of locus coeruleus stimulation on neuronal activities in the rat superior colliculus. Japanese J Physiol 32, 1011–1014.

23.

, Wang

(2018) Modulation of innate defensive responses by locus coeruleus-superior colliculus circuit. J Exp Neurosci 12, 1179069518792035.

24.

Chaturvedi

, Van Gisbergen

JAM

(2000) Stimulation in the rostral pole of monkey superior colliculus:effects on vergence eye movements. Exp Brain Res 132, 72–78.

25.

Aston-Jones

, Chiang

, Alexinsky

(1991) Discharge of noradrenergic locus coeruleus neurons in behaving rats and monkeys suggests a role in vigilance. Prog Brain Res 88, 501–520.

26.

Neves

, van Keulen

, Yang

, Logothetis

, Eschenko

(2018) Locus coeruleus phasic discharge is essential for stimulus-induced gamma oscillations in the prefrontal cortex. J Neurophysiol 119,904–920.

27.

Vazey

, Moorman

, Aston-Jones

(2018) Phasic locus coeruleus activity regulates cortical encoding of salience information. Proc Natl Acad Sci U S A 115, E9439–E9448.

28.

Nieuwenhuis

, Aston-Jones

, Cohen

(2005) Decision making, the P3, and the locus coeruleus–norepinephrine system. Psychol Bull 131, 510–532.

29.

Jodo

, Aston-Jones

(1997) Activation of locus coeruleus by prefrontal cortex is mediated by excitatory amino acid inputs. Brain Res 768, 327–332.

30.

Gamlin

, Yoon

(2000) An area for vergence eye movement in primate frontal cortex. Nature 407, 1003–1007.

31.

Sanchez

, Rowe

(2018) Role of neural integrators in oculomotor systems: A systematic narrative literature review. Acta Ophthalmol 96, e111–e118.

32.

Kozicz

, Bittencourt

, May

, Reiner

, Gamlin

PDR

, Palkovits

, Horn

AKE

, Toledo

CAB

, Ryabinin

(2011) The Edinger-Westphal nucleus: A historical, structural and functional perspective on a dichotomous terminology. J Comp Neurol 519, 1413–1434.

33.

May

, Warren

, Bohlen

, Barnerssoi

, Horn

AKE

(2016) A central mesencephalic reticular formation projection to the Edinger-Westphal nuclei. Brain Struct Funct 221, 4073–4089.

34.

Bohlen

, Warren

, May

(2017) A central mesencephalic reticular formation projection to medial rectus motoneurons supplying singly and multiply innervated extraocular muscle fibers. J Comp Neurol 525, 2000–2018.

35.

Dos Santos Júnior

, Da Silva

KRT

, Haemmerle

CAS

, Batagello

, Da Silva

, Lima

, Da Silva

, Diniz

, Sita

, Elias

, Bittencourt

(2015) The centrally projecting Edinger–Westphal nucleus—I: Efferents in the rat brain. J Chem Neuroanat 68, 22–38.

36.

, Chen

, Pan

, Li

, Pang

, Guo

, Shu

, Cai

, Pei

, Liu

, Afewerky

, Tian

, Zhu

, Lu

(2018) Serotonin receptor 2c-expressing cells in the ventral CA1 control attention via innervation of the Edinger–Westphal nucleus. Nat Neurosci 21, 1239–1250.

37.

Solé Puig

, Pérez Zapata

, Aznar-Casanova

, Supèr

(2013) A role of eye vergence in covert attention. PLoS One 8, e52955.

38.

SoléPuig

, Aznar-Casanova

, Supér

(2013) Difference in visual processing assessed by eye vergence movements. PLoS One 8, e72041.

39.

SoléPuig

, Pérez Zapata

, Puigcerver

, Esperalba Iglesias

, Sanchéz García

, Romeo

, Supèr

(2015) Attention-related eye vergence measured in children with attention deficit hyperactivity disorder. PLoS One 10, e014528.

40.

SoléPuig

, Pérez Zapata

, Puigcerver

, Cañete Crespillo

, Supèr

(2016) Attentional selection accompanied by eye vergence as revealed by event-related brain potentials. PLoS One 1, e0167646.

41.

Solñ Puig

, Romeo

, Cañete Crespillo

, Supèr

(2017) Eye vergence responses during a visual memory task. Neuroreport 28, 123–127.

42.

Esposito

, Sùper

(2018) Vergence responses to face stimuli in young children . . Neuroreport 29, 219–223.

43.

Esposito

, Supèr

(2019) Eye vergence responses to novel and familiar stimuli in young children. Act Psychol 193, 190–196.

44.

Jimñnez

, Avella-Garcia

, Kustow

, Cubbin

, Corrales

, Richarte

, Esposito

, Morata

, Perera

, Varela

, Cañete

, Faraone

, Supèr

, Ramos-Quiroga

(2020) Eye vergence responses during an attention task in adults with ADHD and clinical controls.. J Atten Disord, doi: 10.1177/1087054719897806.

45.

Jimñnez

, Romeo

, Pñrez Zapata

, Solñ Puig

, Bustos Valenzuela

, Cañete

, Varela Casal

, Supèr

(2020) Eye vergence responses in children with and without reading difficulties during a word detection task. Vision Res 169, 6–11.

46.

Feil

, Moser

, Abegg

(2017) The interaction of pupil response with the vergence system. Graefes Arch Clin Exp Ophthalmol 255, 2247–2253.

47.

Van der Wel

, van Steenbergen

(2018) Pupil dilation as an index of effort in cognitive control tasks: A review. Psychon Bull Rev 25, 2005–2015.

48.

Zñnon

(2019) Eye pupil signals information gain . Proc Biol Sci 286, 20191593.

49.

Reimer

, McGinley

, Liu

, Rodenkirch

, Wang

, McCormick

, Tolias

(2016) Pupil fluctuations track rapid changes in adrenergic and cholinergic activity in cortex. Nat Commun 7, 13289.

50.

Foutiou

, Stergiou

, Tsiptsios

, Lithari

, Nakou

, Karlovasitou

(2009) Cholinergic deficiency in Alzheimer’s and Parkinson’s disease: Evaluation with pupillometry. Int J Psychophysiol 73, 143–149.

51.

Joshi

, Li

, Kalwani , Rishi

, Gold , Joshua

(2016) Relationships between pupil diameter and neuronal activity in the locus coeruleus, colliculi, and cingulate cortex. Neuron 89, 221–234.

52.

Folstein

, Folstein

, McHugh

(1975) Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

53.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

54.

Casals-Coll

, Sanchez-Benavides

, Quintana

, Manero

, Rognoni

, Calvo

, Palomo

, Aranciva

, Tamayo

, Peña-Casanova

(2013) Spanish normative studies in young adults (NEURONORMA young adults project): Norms for verbal fluency tests. Neurologia 28, 33–40.

55.

Palomo

, Casals-Coll

, Sánchez-Benavides

, Quintana

, Manero

, Rognoni

, Calvo

, Aranciva

, Tamayo

, Peña-Casanova

(2013) Spanish normative studies in young adults (NEURONORMA young adults’ project): Norms for the Rey-Osterrieth Complex Figure (copy and memory) and Free and Cued Selective Reminding Test. Neurologia 28, 226–235.

56.

Jainta

, Hoormann

, Jaschinski

(2009) Accommodation modulates the individual difference between objective and subjective measures of the final convergence step response. Ophthalmic Physiol Opt 29, 62–172.

57.

Parra

, Ascencio

, Urquina

, Manes

, Ibáñez

(2012) P300 and neuropsychological assessment in mild cognitive impairment and Alzheimer dementia. Front Neurol 3, 172.

58.

Murphy

, Robertson

, Balsters

, O’connell

(2011) Pupillometry and P3 index the locus coeruleus-noradrenergic arousal function in humans. Psychophysiology 48, 1532–1543.

59.

Takeuchi

, Duszkiewicz

, Sonneborn

, Spooner

, Yamasaki

, Watanabe

, Smith

, Fernández

, Deisseroth

, Greene

, Morris

(2016) Locus coeruleus and dopaminergic consolidation of everyday memory. Nature 537, 357–362.

60.

LiuY , RodenkirchC , Moskowitz

, Schriver

, Wang

(2017) Dynamic lateralization of pupil dilation evoked by locus coeruleus activation results from sympathetic, not parasympathetic, contributions. Cell Rep 20, 3099–3112.

61.

Molitor

, Ko

, Ally

(2015) Eye movements in Alzheimer’s disease. J Alzheimers Dis 44, 1–12.

62.

Yang

, Kapoula

(2011) Distinct control of initiation and metrics of memory-guided saccades and vergence by the FEF: A TMS study. PLoS One 6, e20322.

63.

Lehmann

, Corneil

(2016) Transient pupil dilation after subsaccadic microstimulation of primate frontal eye fields. J Neurosci 36, 3765–3776.

64.

Ebitz

, Moore

(2017) Selective modulation of the pupil light reflex by micro-stimulation of prefrontal cortex. J Neurosci 37, 5008–5018.

65.

Mather

, Huang

, Clewett

, Nielsen

, Velasco

, Tu

, Han

, Kennedy

(2020) Isometric exercise facilitates attention to salient events in women via the noradrenergic system. Neuroimage 210, 116560.

66.

Cherng

, Baird

, Chen

, Wang

(2020) Background luminance effects on pupil size associated with emotion and saccade preparation. Sci Rep 10, 15718.

67.

Polich

, Kok

(1995) Cognitive and biological determinants of P300: An integrative review. Biol Psychol 41, 103–146.

68.

Adekani

, Choi

, Hossein-Zadeh

, Porjesz

, Tanabe

, Lim

, Bilder

, Helpern

, Begleiter

(2002) Functional magnetic resonance imaging of brain activity in the visual oddball task. Cogn Brain Res 14, 347–356.

69.

Lastra

, Morales-Rodriguez

, Penzol-Diaz

(2001) Neurophysiological study and use of P300 evoked potentials for investigation in the diagnosis and of follow-up of patients with Alzheimer’s disease. Rev Neurol 32, 525–528.

70.

Lai

, Lin

, Liou

, Liu

(2010) The role of event-related potentials in cognitive decline in Alzheimer’s disease. Clin Neurophysiol 121, 194–199.

71.

Polich

(1989) P300 and Alzheimer’s disease. Biomed Pharmacother 43, 493–499.

72.

Hedges

, Janis

, Mickelson

, Keith

, Bennett

, Brown

(2016) P300 amplitude in Alzheimer’s disease: A meta-analysis and meta-regression. Neurosci 47, 48–55.

73.

Koronyo-Hamaoui

, Koronyo

, Ljubimov

, Miller

, Ko

, Black

, Schwartz

, Farkas

(2010) Identification of amyloid plaques in retinas from Alzheimer’s patients and noninvasive in vivo optical imaging of retinal plaques in a mouse model. Neuroimage 54, S204–S217.

74.

Liao

, Zhu

, Peng

(2018) Potential utility of retinal imaging for Alzheimer’s disease: A review. Front Aging Neurosci 10, 188.

75.

Crawford

, Devereaux

, Higham

, Kelly

(2015) The disengagement of visual attention in Alzheimer’s disease: A longitudinal eye-tracking study. Front Aging Neurosci 7, 118.

76.

Giorgi

, Ryskalin

, Ruffoli

, Biagioni

, Limanaqi

, Ferrucci

, Busceti

, Bonuccelli

, Fornai

(2017) The neuroanatomy of the reticular nucleus locus coeruleus in Alzheimer’s disease. Front Neuroanat 11, 80.

77.

Dahl

, Mather

, Düzel

, Bodammer

, Lindenberger

, Kühn

, Werkle-Bergner

(2019) Rostral locus coeruleus integrity is associated with better memory performance in older adults. Nat Hum Behav 3, 1203–1214.

78.

Kret

, Sjak-Shie

(2019) Preprocessing pupil size data: Guidelines and code. Behav Res Methods 51, 1336–1342.

79.

Brooks

, Zoumpoulaki

, Bowman

(2016) Data-driven region-of-interest selection without inflating Type I error rate. Psychophysiology 54, 100–113.