Comparison of effects between SMR/delta-ratio and beta1/theta-ratio neurofeedback training for older adults with Mild Cognitive Impairment: a protocol for a randomized controlled trial

Since improvement in algorithms in electroencephalography (EEG) offers a promising approach to developing a database for assessing MCI and AD subjects, several researchers have underscored links between EEG signal and AD/MCI. With age, fast rhythms have been found to decrease [5, 6]. Most commonly in the healthy aging process, EEG changes are not consistently observable but might involve a slight decrease in frequency and amplitude in the alpha band (8–12 Hz), an increase in delta (1–4 Hz) and theta (4–8 Hz) activities unlike young healthy subjects [7, 8]. In AD, theta and delta rhythms are abnormally more dominant in temporal regions of the brain than in other cerebral regions and posterior alpha rhythms significantly decrease [9‐13]. Other promising studies in pathophysiology revealed that patients with MCI showed dominant theta activity in the parietal and temporal areas [14‐16]. Moretti et al. showed in numerous studies that the abnormal increase in the power ratio alpha3/alpha2 in temporo-parietal brain regions is correlated with hippocampal atrophy in MCI and AD patients [17‐19]. Amnestic Mild Cognitive Impairment (A-MCI) subjects present decreased power of the posterior alpha band correlated with cognitive decline [12, 20‐22]. Dubois et al. emphasized the decrease in the theta/alpha ratio explained by an increase in alpha oscillations over time in prefrontal areas in people who have Aβ42 deposition and people with prodromal AD. Then, other studies focused on the analysis of the beta band (13–30 Hz) changes at the MCI stage [15]. In recent years, progress in the understanding of MCI and AD pathophysiology has opened several therapeutic perspectives and facilitated the design of more adequate rehabilitation programs [23‐27]. Interest in brain-training programs based on sustained cognitive exercises is rising [28, 29]. Non-pharmacological cognitive interventions have been recognized as an efficient method of enhancing cognitive and functional abilities of MCI [30‐32].

Neurofeedback (NF), also called electroencephalography Biofeedback (EEG Biofeedback) is one of the promising techniques that received international accreditation boards’ and treatment recommendations of evidence level A from the American Academy of Pediatrics for attention-deficit hyperactivity disorder (ADHD). Other promising results have been found for therapeutic efficacy in autism [33], patients with pharmaco-resistant epilepsy [34], and traumatic brain injury [35]. Neurofeedback uses real-time displays of brain activity—most commonly EEG—to teach self-regulation of brain function. This technique, defined as a closed-loop application, helps individuals to control or modify their cortical activity through learned self-regulation. Precisely, it is operant behavioral training that aims at improving the cognitive functions and that affects regulation by modulating brain electrical activity. The patient receives feedback of acoustic or/and visual signals (graphic or video games/animations/movies) when their rhythmic cortical electrical activity of specific cortical areas has exceeded an upper threshold [36]. The subject has to improve the quality of the signal by a continuing focus in an absolute relaxing state and not exceed a determined upper threshold in a frequency band. The aims are to obtain better alertness and eliminate symptoms of anxiety, consequently inducing better behavior.

Among the protocols of NF training, the sensorimotor brain rhythm (SMR, 12–15 Hz) is an oscillatory rhythm recorded over central scalp regions, generated in a reticular-thalamo-cortical network [37, 38]. SMR/delta NF training involves increasing the synaptic strengths and sensitivity within this network while maintaining low delta activity to avoid sleep (delta, 0.5–4 Hz) [39]. It was suggested that SMR NF training might facilitate thalamic inhibitory mechanisms and block motor activity that interferes with information processing [37]. In accordance with this assumption, few studies found memory and attention improvements after SMR NF training [36, 40‐42].

The beta1/theta protocol consists of beta1 dominance (12–16 Hz band) and reduction of theta oscillations (4–7 Hz band) in the central frontal area. Beta1 is described as responsible for logical thinking, concentration, memory and emotional status [43]. Beta2 can be sign of anxiousness; NF training within this range might be conducted with prudence. Theta rhythm is associated with neurological and psychological functions in the limbic system, it serves regulatory control of arousal, affective and mental states [44]. Its excessive presence in frontal areas while the person is awake, indicates a decrease in vigilance with concentration and attention disorders. Thus, an increase in the central frontal beta1 band and a decrease in the central frontal theta band can be linked with an increase in vigilance and concentration and a decrease in diurnal sleepiness [43, 45].

Recently, several studies focused on investigating the benefits of NF training on cognitive functions in elderly subjects. It was shown that elderly patients, who completed a NF training program after stroke to generate SMR modulation on central regions of brain, improved in declarative learning memory [40, 46‐50]. Other studies demonstrated increasing performance in working memory in healthy aging with a theta/alpha protocol training on Fz [29, 51] or with an upper alpha training in central regions of the brain [52]. Luijmes et al. [53] observed an increase in learning memory, recognition and recall of information after NF treatment in patients with AD. From a recent pilot study conducted by our research team at the Broca University Hospital in Paris, we investigated the effects of SMR/theta NF training on cognitive performance in elderly patients with MCI. Results showed that cognitive profits were minor, but anxiety decreased and quality of life and sleep improved. EEG data analyzed after the training program indicated that slow frequency bands decreased while alpha and sensorimotor rhythms were more dominant than in pre-training. Furthermore, this study enabled us to identify a few limits in the EEG NF application for older adults with MCI. These preliminary results allowed us to suggest some recommendations to avoid them such as the need to adapt the NF tasks, give explicit information about NF technique and all supplies, and to adjust the threshold to the cognitive and specific profile of the patient.

9pt?>Neurofeedback training seems to be a promising aproach in patients with AD or neurological disorders if this technique is adapted to elderly patients with MCI. On this basis, it would be relevant to further this technique in subjects at risk of developing AD. Based on the studies of Reichert et al. [47], Kober et al. [40] and Luijmes et al. [53], our objective is to examine: (1) the effects of SMR/delta NF training and the beta1/theta-ratio protocols on attention and memory performances in subjects with MCI, (2) to verify whether MCI patients exhibit a decrease or an increase in EEG activities in a resting state after SMR/delta NF training and/or alpha/theta-ratio NF training and (3) to assess the effects of the two training protocols on other cognitive and mood measures.

We hypothesize that (1) central frontal beta1/theta-ratio NF training might be effective in improving attention capacities whereas the SMR/delta protocol in Cz might be effective in improving memory performances and (2) changes in EEG patterns in post-SMR/delta-ratio and beta1/theta-ratio NF training would be observed in the intervention group but not in the control group. This paper has been conceived under consideration of the spirit guidelines (Additional file 1).

This study is a prospective, randomized controlled and single-blinded clinical trial as shown in Fig. 1. This study will be reported in accordance with both the Consolidated Standards of Reporting Trials (CONSORT) Statement, the CONSORT Statement for non-pharmacological treatment interventions and respecting Neurofeedback Protocol Guidelines [45]. One of the main investigators will inform patients, verbally and in a written document, of the objectives of the study, the project progress with EEG and neuropsychological assessments, and the potential risks. Each participant will be asked to sign a written consent form to ensure their voluntarily participation along with the information that they have right to discontinue the project any time without penalty and their agreement for the use of collected data. Signed consent forms will be kept by the main investigators.

After screening and obtaining their written consent, participants who are deemed eligible for the study will be enrolled and the randomization schedule will be designed to yield an assignment ratio of 1:1:1 between the three arms. Randomization will be centralized (CleanWEB; Telemedicine Technologies) and not stratified by blocks of variable size. No one from any group will know if they belong either to an experimental group or the control group (Fig. 1).

The sample size calculation was estimated on the basis of our feasibility study evaluating the effects of NF training on cognitive performances in MCI subjects (registered at ClinicalTrials.gov under the identifier: NCT03526692, on 16 May 2018). In an analysis of covariance study (ANCOVA), sample sizes of 20 are obtained for each of the three groups whose expected averages of TMT B are 140 in the control group and 105 s in two intervention groups with a standard deviation of 45. The supposed R-squared of the model is 0.20. The total sample of 60 subjects has a power of 80% to detect differences between means using an F test with a significance level of 0.05.

Sixty older adults with MCI will be recruited by the geriatric medicine physicians through the outpatient memory service from Broca University Hospital in Paris. All participants will be eligible if they meet MCI criteria according to Petersen et al. [54]; are aged 60 years or over; have a Mini Mental State Examination (MMSE) [55] score > 24; a subjective memory complaint confirmed by an informant; perform at/or below 1.5 standard deviations from the mean for age and education-matched norms on more than one of the neuropsychological tests; preserve activity of daily living; and absence of dementia. The exclusion criteria will include patients with epilepsy, stroke, tumor, head trauma, personality disorders, and psychiatric disorders with the exception of mild depression using the GDS (Geriatric Depression Scale, normative score: normal 0–9, mildly depressed 10–19, very depressed 20–30) [56]. Participation in other trials will be prohibited during the study.

In order to guarantee blinded treatment allocation, an independent statistician engineer, not engaged in this investigation, will carry out randomization. Applying equal allocation ratios to all three conditions, the process will be stratified. Statistical analyses will be conducted using the conservative intent-to-treat population in order to avoid statistical bias. Primary and secondary outcomes, and EEG recordings will be assessed by trained investigators and technicians who will be blind to the patient’s study status.

Effect sizes and confidence intervals will be calculated as Standardized Mean Change Index for within-group comparisons and as Standardized Mean Index difference for between-group comparisons and corrected for small sample sizes.

Technical specifications of the trial database, such as variable names, age, disease, frequency of visits and sessions will be predefined and documented in the database manual and in the electronic Case Report Form. All information on the Case Report Forms will be confidential and traceable to source documents in the patient’s file; the original documents will be kept by the main investigators of the study at the Broca University Hospital in Paris. At the end of the study, personal information will be deleted. The final trial dataset and disclosure of contractual agreements will be kept at the Broca University Hospital in Paris. Interim analyses on the acquired data will be conducted and presented in meetings and conferences. Results of the final analyses will be presented at scientific conferences and published in scientific journals.

Analysis will be performed based on an intent-to-treat (ITT) principle; ITT analysis includes every subject who is randomized according to randomized treatment assignment. Baseline characteristics of the treatment groups will be summarized descriptively without any statistical tests of significance. Categorical variables will be described as numbers and percentages and continuous variables as means with standard deviation.

No interim analysis and no subgroup analysis were planned. All tests will be two-sided at a 0.2 significance level. All statistical analyses will be carried out using R statistical software version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria).

EEG data will be processed with MATLAB R2011b, MathWorks Inc., Natick, MA, USA.

We will calculate the power spectral density of EEG signals to extract the relative power (RP) and absolute power (AP) for each frequency band. By definition, the RP is a value that compares the AP within a frequency band with the total AP. So, relative theta, relative alpha, relative SMR and relative beta will be calculated in the central, frontal and temporal areas during the rest ECB and rest EOB conditions in pre-NF and post-NF training. The ANCOVA model will provide the differences in RP between these two times. When the difference is positive, it means that the RP in post-NF training for that channel is larger than the RP in pre-NF training. The variables will be used to assess differences between the SMR NF group and the control group.

The time schedule and analysis plan are illustrated in Table 2.

To ensure that a large number of participants finish the study and participate in as many sessions as possible with motivation, the PhD candidate specialized in electrophysiology, neuropsychology and gerontology will meet each participant to explain specific program information, show the equipment (EEG, and EEG NF), and give a demonstration of EEG NF training before assessment at baseline. All participants will receive times and contact details of study investigators if they have questions or encounter difficulties during the intervention program.

An independent technician specialized in EEG will record EEG in each patient. A neuropsychologist with full knowledge of all measurement scales and administration tests will ensure assessments and data collection. Neurofeedback training sessions will be ensured by a PhD candidate specialized in electrophysiology, neuropsychology and gerontology. In order to guarantee blinded treatment allocation, an independent engineer in biostatistics, not engaged in this investigation, will carry out statistical analysis and EEG analysis.

The Broca University Hospital in Paris will provide access to neuropsychological tests, questionnaires, materials, computers and infrastructure that are necessary for conducting the study. Neurofeedback software and license EEGDigiTrack Biofeedback plus module and the multichannel amplifier EEGDigiTrack SimplEEG32 are provided by the company Inc. Elmiko Medical, Warsaw, Poland.