Brain Functional Changes Accompany Modulatory Effects of Transcranial Direct Current Stimulation in Cognitive Impairment

As apriori registration was not done before the initiation of the study and all the participants already underwent tDCS intervention, the investigators are herewith submitting the protocol of the study retrospectively following the completion of the project. This study was conducted at the National Institute of Mental Health and Neurosciences (NIMHANS), Bengaluru, India, with the approval of the NIMHANS Institutional Ethics Committee.

Study Sample: All consenting participants of both sexes (age range: 55 - 84 years) meeting the inclusion and exclusion criteria (see below), who consecutively attended the outpatient services of the Geriatric Clinic and Services (GCS), NIMHANS from August 2019 to December 2022 with memory complaints were recruited in this study with their signed informed consent. Only right-handed individuals were recruited as assessed using the Edinburgh Handedness Inventory (EHI). Participants were screened to ensure that only those with optimal hearing and visual acuity were recruited. The clinical diagnoses of MCI and mild AD made by clinicians at the GCS at NIMHANS were confirmed using the National Institute on Ageing - Alzheimer's Association (NIA-AA) diagnostic criteria and Clinical Dementia Rating [CDR] (CDR scores of 0.5 for MCI and 1 for mild AD). Participants with CDR scores above 1 were excluded from this study. Other exclusion criteria included concurrent medical conditions like hypothyroidism, hypercalcemia, vitamin B12 deficiency, uncontrolled hypertension and diabetes, neurosyphilis, normal pressure hydrocephalus, subdural hematoma, etc., requiring initiation of medical/surgical interventions and/or titration of medications; pre-existing major psychiatric and neurological illnesses, such as schizophrenia spectrum disorders, bipolar affective disorder, major depressive disorder, obsessive-compulsive disorder, substance dependence, intellectual disability disorder, Parkinson's disease and related disorders, stroke, epilepsy, and other chronic neurological/neurodegenerative disorders, and significant head injury; and current usage of antipsychotics, antidepressants, and benzodiazepines. Forty patients (19 MCI and 21 mild AD) consented to participate in the study and completed the tDCS intervention protocol, which comprised daily 20-minute tDCS sessions for 10 days. All participants underwent detailed neuropsychological assessments, resting-state functional magnetic resonance imaging (rsfMRI), task-based functional magnetic resonance imaging (tbfMRI) as well as resting state magnetoencephalography (rsMEG) following tDCS intervention.

Neuropsychological Battery for Elderly (NNB-E): This battery includes tests of episodic memory, executive functions, attention, visuospatial function, and parietal focal signs and required 1.5 hours to administer and score. Assessments were carried out at baseline (pre-tDCS) and after completion of the last tDCS session (post-tDCS). These pre- and post-tDCS assessments were conducted in a way that prevented the same researcher from doing both tests for a particular participant. Additionally, the person conducting the post-tDCS assessment was blind to the results of the pre-assessment. Two different forms were used for the pre-and post-tDCS assessments of a participant to avoid practice effects. Shapiro Wilk test was performed to check the normality distribution of the differences between pre- and post-tDCS scores for each sub-test (n=26) of NNB-E. Accordingly, paired t-tests and Wilcoxon signed rank tests were carried out respectively for normally and non-normally distributed sub-test scores to look for significant effects of tDCS intervention.

Magnetic Resonance Imaging (MRI): The MRIs were acquired on a 3Tesla Philips Ingenia CX (Philips Healthcare, Netherlands) scanner with a 32-channel phased-array head coil. During rs-fMRI acquisition, the participants were instructed to remain physically and mentally relaxed with their eyes open and not focus on anything particular. These functional MR images corresponding to each participant were visually examined for scanner or acquisition-related artifacts. The rs-fMRI data of one participant each from the MCI and mild AD groups showed more than 5% intensity variation in two consecutive volumes indicating poor quality of functional images due to motion artifacts validated using Derivative of root mean square VARiance over voxelS (DVARS). After the exclusion of both pre- and post-rsfMRI scans of these two participants, the data of the remaining 38 participants (n=18 in the MCI group and n=20 in the mild AD group) were further analyzed. The seed for the seed-to-voxel resting state functional connectivity (S2V-rsFC) analysis was set using the Harvard cortical atlas, as the left middle frontal gyrus (lMFG), corresponding to the left DLPFC, the site of the anodal tDCS. The analysis between pre- and post-tDCS intervention was performed using the spherical mask of lMFG as 'seed' and the whole-brain volume for 'voxels'. Two-tailed parametric statistics were implemented to obtain the results using a cluster-level extent threshold (p-FDR corrected) with voxel threshold at p \< 0.001 (p-uncorrected) and cluster threshold at p \< 0.05 (p-corrected).

A novel episodic memory paradigm, comprising two tasks: an incidental encoding task and an intentional retrieval task, was used for tb-fMRI acquisition. In the incidental encoding task, the participants were instructed on each trial to indicate using a button press whether the presented visual images were of living or non-living objects. In the intentional retrieval task, the instruction for each trial was to indicate using a button press, whether the presented visual images were seen before (memory images, MI) or not seen before (non-memory images, nMI). The optimal number of trials for the incidental encoding task (living objects: 55; non-living objects: 55; fixation cross: 56) and the intentional retrieval task (memory images, MI: 55; non-memory images, nMI: 55; fixation cross: 56) was estimated using the fMRI stimulator to maximize the predictable variance between the three trial classes for each of the two tasks. Only the participants showing more than 60% performance accuracy on both incidental encoding and intentional retrieval were analyzed. Task-based fMRI data pre-processing and analyses of remaining participants with early AD (n=30; MCI=15, mild AD=15) were carried out using FMRI Expert Analysis Tool (FEAT) Version 6.00 of FMRIB Software Library (FSL), implementing high-quality model-based fMRI data analysis. Z (Gaussianised T/F) statistical images were thresholded non-parametrically using clusters determined by Z > 2.3 and a corrected cluster significance threshold of P=0.05.

Magnetoencephalography (MEG): The pre- and post- rsMEG data were acquired using the Elekta Neuromag® TRIUX™ MEG Scanner. Two participants (1 in the MCI sample and 1 in the mild AD sample) opted out of taking part in MEG acquisition, and four participants (3 in the MCI sample and 1 in the mild AD sample) did not participate in post-MEG scans; MEG data of the remaining participants with early AD (n=34; MCI=15, mild AD=19) were further preprocessed. The resting eyes closed MEG recordings were carried out in a sitting position employing the following protocol: sampling rate 2000 Hz, empty room recording=2 minutes, resting-state MEG=15 minutes. EEG data were acquired concurrently to ensure that the participants were alert and relaxed during the recording. Electrocardiogram (ECG), electromyography (EMG), and electrooculogram (EOG) were acquired to remove the heart, muscle, and ocular artifacts, respectively. Quality control (QC) of the MEG data was done using MaxfilterTM (ver 2.2), a built-in software for the ELEKTA/Neuromag system. Suppression of interference due to environmental and periodic artifacts was carried out using the spatiotemporal signal space separation (tSSS) method. Using the Welch method, a power spectrum analysis was performed on 5 minutes of artifact-free MEG recording. Extracted power spectrum density (PSD) values for neural oscillations covered canonical frequency bands [delta (2-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta (13-30 Hz), gamma (30-60 Hz) and gamma2 (60-90 Hz)]. A paired T-test was performed between the pre- and post-tDCS values of normalized PSD, correcting for multiple [permutation testing (1000 randomizations), Monte Carlo method] comparisons (FDR-corrected), controlling for time and frequency.

The dependence between the phase of low-frequency oscillations (fP) and the amplitude of high-frequency rhythms (fA) was quantified using phase amplitude coupling (PAC). The coupling strength for each participant was estimated using time-resolved PAC (tPAC), which shows the least relative error. The tPAC was estimated by searching for the fP oscillation with the strongest PAC to fA bursts, over a time window, which slides on the input MEG data. A 10-second epoch length was used for analysis, and since the fP band of interest was between 4-8 Hz, three full cycles of the fP band were chosen to produce a sliding window length of 0.75 seconds. The average comodulograms were extracted using tPAC maps of all the patients with MCI (n=15) and mild AD (n=19) in the left entorhinal cortex which plays a critical role in episodic memory and is vulnerable to early pathophysiology in AD. A paired t-test was performed outside this software using MATLAB® (https://www.mathworks.com) to find out the significant (FDR-corrected p \< 0.05) difference in coupling strength between phase of low frequencies (4-8 Hz) and amplitude of high frequencies (30-90 Hz) correcting for multiple comparisons across time and frequency.