The goal of this one-arm clinical trial is to determine whether participants with mild cognitive impairment (MCI) can successfully navigate a virtual reality (VR) maze. The VR maze is designed as a training tool aimed at improving participants' spatial navigation abilities.

Main Aims:

1. To determine whether at least 70% of older adults enrolled in the study can complete twenty-four 50-minute training sessions over a 4-month period.
2. To assess whether combining virtual reality with EEG recordings can be used to measure brain activation and changes in brain activation associated with spatial navigation learning.

Participants will:

1. Walk in an open, unobstructed space while wearing VR goggles.
2. Explore up to fifty different virtual mazes in sequence and attempt to find their way through each one.

Specific Aims: With no cure available yet, lifestyle interventions to delay onset of Alzheimer's disease and related dementias are essential. Up to a third of dementia cases may be preventable by engaging in protective behaviors, such as staying cognitively active, according to observational data. Yet, cognitive training protocols to delay onset often fall short. The investigator team hypothesizes that designing an intervention informed by specific measurements from brain regions subserving cognition will yield better results.

This pilot project is focused on spatial navigation (SN), the ability to travel familiar/unfamiliar environments, which is particularly suited as a target for early intervention. Difficulties forming new and maintaining old spatial memories is a common and early sign of Alzheimer's disease and can lead to disorientation and dependence in performing daily activities. Tau and amyloid-beta accumulation starts in regions subserving SN, such as mediotemporal and posterior parietal cortex. Even though SN difficulties present an important target, there are few clinical trials aimed at SN. Many trials are desktop- or virtual reality-based with learners using a handheld device to train SN. A major limitation of these training methods is that it deprives learners from movement-related sensory and kinematic information during active navigation.

To overcome this limit, a full-immersive virtual-reality (VR) maze where participants train their navigation abilities has been developed. Importantly, the maze design is informed by specific contributions of mediotemporal, posterior parietal cortex, and retrosplenial complex in SN. To quantify regions-specific contributions, the investigator team applied Mobile Brain Body Imaging (MoBI) to synchronously record body movement with EEG to record and analyze brain activity at the source level during active movement through space. The maze is designed to break down learning into two discrete periods: 1) Stand/Encode the maze from a birds-eyes view to induce allocentric spatial coding (i.e., independent from navigator's position/orientation) and 2) Walk/Navigate the maze with walls raised so that only corridors/intersections are visible during walking to induce egocentric spatial coding (i.e., navigator dependent; turn left at the next intersection). Evidence suggests that the mediotemporal cortex is involved in allocentric and posterior parietal cortex in egocentric spatial coding, with the retrosplenial complex integrating across allo- and egocentric mental frames. Using MoBI, the study team will identify and track intervention-related changes in brain activation during allo- and egocentric spatial coding. 30 community-dwelling non-demented older adults will be enrolled for a personalized supervised one-armed clinical trial SN intervention composed of 24 sessions. VR mazes gradually increase in complexity based on individual learners' progress. Dementia-at-risk status will be determined with cut-scores on validated telephone-based screeners.

Aim 1: Determine feasibility of a personalized Virtual Reality-Spatial Navigation (VR-SN) maze training protocol. The feasibility of the VR-SN maze protocol to train non-demented older adults will be examined. Feasibility metrics, including recruitment sources, implementation (i.e., intervention session completion rates), retention rates, acceptability (post-study interviews) and safety (adverse events monitoring) will be assessed using a mixture of qualitative and quantitative assessment tools.

Aim 2: Explore training-induced neuroplasticity in older adults participating in VR-SN exercises. Validated Neuroplasticity behavioral (Floor Maze test, FMT) and neurophysiological (MoBI) measures are described in the Outcome Measure sections. The FMT is a 7-by-10-foot maze taped on the floor and measures the time needed to navigate the maze, which the study team validated as a robust predictor of cognitive impairment and studies have linked to Alzheimer pathology. Neurophysiological outcomes are based on prior VR-SN MoBI measures obtained in young adults. Hypothesis 2a: VR-SN training will improve performance on the FMT. Hypothesis 2b: VR-SN training will increase modulations of retrosplenial theta power during Stand/Encode and posterior parietal alpha power during Walk/Navigate periods. Hypothesis 2c: Increased modulation in theta/alpha power will be correlated with improvement in the Floor maze test.

VR is a One-Size-Fits-One approach: VR has proven to be effective in helping patients with gross motor and balance difficulties, post-traumatic stress disorder, and cognitive impairment. The versatility of VR allows for generating diverse sets of controlled/safe scenarios designed specifically to a patient's needs. This study's VR SN approach facilitates sustained learning over time through novelty (i.e., new mazes can be created for each session), appropriate challenge (i.e., maze complexity can be matched to learner's progress), and positive feedback (i.e., reaching the maze goal). Participants often report enjoying the game-like maze exercises which is essential to sustain motivation. The investigator team argues that this VR protocol with a focus on spatial navigation may prove clinically significant, much like prior VR interventions that yielded higher gains compared to non-VR interventions. Providing learners the opportunity to actively move and navigate in virtual space (VR SN) while recording related brain activation (VR SN MoBI) will increase scientific rigor and address prior shortcomings (e.g., in-place navigation). Compared to alternative imaging approaches (e.g., MRI, PET), EEG is low-tech, fully portable, and cost-efficient. Ultimately, this may justify an ambulatory EEG protocol commonly used in clinical settings.

Innovation: SN engages spatial (e.g., allo/egocentric spatial coding) and non-spatial cognitive (e.g., attention) abilities. Furthermore, SN requires integrating and flexible switching between allocentric and egocentric mental frames. If the innovative and novel VR-SN MoBI maze protocol is developed and successfully determines region-specific brain activation associated with allo- and egocentric encoding, this will allow the complex interplay of cognitive processes that lead to disorientation and the inability to navigate familiar and unfamiliar environments to be unfolded. Differentiating and recognizing specific contributions, can lead to individualized targeted training protocols to tackle spatial and non-spatial processes, as well as issues of spatial coding related to switching/integrating allocentric and egocentric mental frames. Tracking pre/post intervention changes in region-specific brain activation of SN will provide novel candidate markers of neuroplasticity that can be tested and used as quantifiers (e.g., duration, frequency, complexity) to develop better intervention against cognitive decline and dementia.

Neuroplasticity based on the assumption that engagement in challenging mental exercises (e.g., SN) confers greater protection to respective brain regions (e.g., medial temporal cortex) and/or leads to the recruitment of novel brain regions during task performance, the study team seeks to probe intervention-related changes in EEG brain activation as a means to quantify neuroplasticity. Prior studies have yielded some promising results. For example, in stroke patients recruitment of homologous brain regions is related to recovery of sensorimotor abilities. In contrast, EEG studies probing cognitive/physical interventions and related plasticity have yielded mixed results. EEG spectral power shifts from high to lower frequency bands during dementia have been reported with a reversal in this EEG shift as well as improved cognition after 6 weeks of exercise in patients with MCI. At the same time, studies report no evidence linking cognitive training gains to EEG markers. This proposal builds and advances prior studies by probing for region- and spectral-specific brain activation within a One-Size-Fits-One SN exercise approach to quantify neuroplasticity.

APPROACH Study procedures/assessments

• Telephone Memory Impairment Screen (MIS): The MIS is a brief four-item dementia screen that assesses perceived and recent changes in memory (cut score ≤ 5, sensitivity 85%, specificity 86%).
• 8-item Dementia Screening Interview (AD8): The AD8 is an 8-item dementia screen testing memory, orientation, judgment, and daily function (cut score ≥ 1, sensitivity 74%, specificity 86%).
• Telephone-based Montreal Cognitive Assessment (T-MoCA): The T-MoCA tests different cognitive domains including attention, executive functions, memory, language, and visuospatial/constructional skills and demonstrates sufficient psychometric properties as a screen for mild cognitive impairment (cut score > 17, sensitivity 72%, specificity 59%).
• Floor Maze Test: FMT is an ecologically valid test of SN developed by the investigator team and has been shown to predict MCI as well as correlate with Alzheimer pathology. Participants will be positioned at the entry point and instructed to find their way to the exit point. A fixed 15-second planning period will be given to plan the route. The time elapsed from the end of the planning period to successful exit will be recorded (immediate maze time [IMT], seconds).
• MoBI: navigation-related modulations in region-specific theta and alpha power during Stand/Encode and Walk/Navigate periods using block- and event-related (heel-strike) spectral analysis approaches as detailed in prior MoBI work.

Feasibility domains and benchmarks for success. Aim 1 seeks to test the feasibility of methods and procedures for later use on a larger scale. The study team proposes a 10-day rotational schedule to train thirty older adults participating in 24 VR SN maze sessions over a period of 4 months. A staggered protocol will be applied, with 15 older adults (2 sessions within 10 days) trained in months 6-9 and 15 older adults trained in months 10-14 of this 2-year proposal. Primary criteria for determining success: 1) at least 60% of all eligible older adults can be recruited and 2) 70% of older adults can be trained with the rotational 10-day program (i.e., 378 sessions over 12 months). In addition, data using the classification proposed by Thabane et.al. (see References section) will be collected. Mixed methods using quantitative and qualitative measures to assess feasibility will be applied.

Sample, recruitment, and screening. A random sample of older adults (≥65 of age) will be identified through various means: information materials and presentation at the Fort Washington Senior Center and lists of volunteers that participated in previous studies at Albert Einstein College of Medicine (Division of Cognitive and Motor Aging). Those expressing interest will be screened over the phone using 'dementia-at-risk' cut-scores on either the MIS (>3 to ≤6) and the AD8 (≥1). Participants enrolled will have a spectrum of cognitive function ranging from normal to Mild Cognitive Impairment (T-MoCA ≤ 17). This will be accounted for in multiple ways. 1. Adjusting for MCI status in analysis; 2. Adjusting for T-MoCA score; 3. Conducting stratified analysis by MCI status.

Intervention Protocol VR maze session 1: Start with an individualized, face-to-face introductory session to describe and answer questions about the protocol. Participants will test the VR goggles, EEG cap, and become familiarized with the VR maze task. VR-SN MoBI session 2 (baseline): Each session lasts 50 minutes. Initial maze complexity will be set at the lowest level (i.e., one turn to reach target). Navigation targets can be dropped anywhere within the maze to manipulate navigation complexity in accordance with learner's needs, resources, and progress. A maze will be repeated until performed without errors after which a new maze is introduced. Many VR mazes are pre-programmed to ensure that exercise is feasible, novel and challenging to incentivize long-term SN training. An up-down transformed rule (UDTR) will be used to adjust complexity based on a participant's performance. A three-up/one-down rule, meaning that for three consecutive error-free mazes will be used where the complexity will be adjusted by introducing an additional turn-to-target and for any error the number of turns reduced to reach the target by one. VR maze sessions 3-23 (no MoBI): Participants take part in two sessions (50 minutes, including breaks) within 10 days over a 4-month period. VR-SN MoBI sessions 12 & 24: Participants are trained on the VR maze while MoBI is recorded.

Safety Participation in immersive virtual reality can result in nausea and dizziness in some individuals. To ensure safety, the research assistant will be at the participant's side to spot participants as they ambulate through empty space to ensure safety. Breaks after each learned maze are encouraged. Continued exposure to VR will be limited to a maximum of 10 minutes to allow participants to re-engage with the physical world. A recent meta-analysis in patients with MCI/dementia participating in VR interventions reported no significant increases of adverse effects for VR protocols. All interventions are conducted in medical facilities.

EXPERIMENTAL APPROACH

Details regarding Aim 1 are provided for reference only, to ensure a comprehensive account of the study. The registration of this proposal in ClinicalTrials.gov pertains exclusively to Aim 2.

Aim 1: Determine feasibility of the VR-SN maze training Rationale: Gathering preliminary results about the feasibility of VR-SN exercises for non-demented older adults will help optimize the protocol for a future randomized clinical trial.

Aim 2: Explore training-induced neuroplasticity in older adults following a VR-SN maze training Rationale: Hypotheses are VR-SN maze training reduces the time required to navigate the Floor maze and increases modulations of theta/alpha power; Floor maze time is correlated with theta/alpha power change.

Inclusion Criteria:

• Age 65 and older with amnestic mild cognitive impairment (aMCI);
• Can speak English;
• Agrees to MoBI recording;
• Normal or corrected-to-normal vision/audition;
• Able to walk unassisted for 10 minutes;
• Plan to be in the area for next year

Exclusion Criteria:

• Dementia (Memory Impairment/AD8 screen);
• Medical conditions that affect participation such as vertigo and neck pain;
• Hospitalization in the past six months or plans for surgery affecting participation in the next four months;
• Mobility limitations solely due to musculoskeletal limitation or pain;
• Terminal illness with life expectancy less than 12 months;
• Presence of clinical disorders that overtly alter attention like delirium;
• Active psychoses or psychiatric symptoms;
• Living in nursing home;
• Participation in intervention trial;
• Standard contraindications to EEG including seizure medication, epilepsy, stroke, traumatic brain injury;
• Pregnant women

Demographic, cognitive, clinical, and 32-channel EEG electrophysiological data will be acquired. Virtual reality is presented through SteamVR's HTC Vive technology, specialized for full-immersive 3D experiences using motion tracking to move and interaction within the virtual environment. All data will be de-identified prior to receipt by the repository, but the information needed to generate a global unique identifier for the NIMH Data Archive (NDA) will be collected. Scientific data is expected to reach 10 terabytes.

The clinical, cognitive and EEG data will be analyzed with custom matlab code written using MATLAB® environment and EEGLAB package for EEG data. While MATLAB is commercial software, most universities have site licenses available. All code will be shared on the GitHub lab website. The code can be found by searching for "labname" on GitHub. The main readme.md file for the project will also include instructions.

The research community will have access to data when the award ends. As required by NDA, studies will also be created that contain the data used for every publication. Those studies will be shared when the pre-print is available. NDA studies have digital object identifiers (DOI) to aid in findability. This DOI will be included in relevant publications. NDA will make decisions about how long to preserve the data, but that data archive has not deleted any deposited data up to now. Data will be share before the end of the project or at the time of publication whatever comes first. Data will be stored for at least three years.

Data will be findable for the research community through the NDA Collection that will be established when this application is funded. For all publications, an NDA study will be created. Each of those studies is assigned a digital object identifier (DOI). This data DOI will be referenced in the publication to allow the research community easy access to the exact data used in the publication.