Gamma oscillations are rhythmic changes of the brain's electrical activity in a broad band of frequencies that range between 25 and 135 Hz; these oscillations can be subdivided into slow gamma (25–60 Hz) and fast gamma (60–135Hz) oscillations (1). Gamma oscillations have been found to be involved in several cognitive processes, such as sensory integration, selective attention, and retrieval of memories (2); because of this, several neurological disorders, including mild cognitive impairment (MCI) (3) and Alzheimer's disease (AD) (4), have shown aberrant behavior in the gamma band of the oscillatory activity of the brain (1, 4).

MCI is a progressive disorder diagnosed when an individual presents loss of cognitive abilities beyond what would be expected given his or her age and educational background, but it does not cause functional impairment (5). AD is a progressive disorder that starts as amnestic MCI that gradually evolves into dementia due to progressive neuronal degeneration and loss, which produces an eventual destruction of cognition, personality, and the ability to function independently even in daily life tasks (6).

AD animal models have shown a reduced gamma power in several brain areas such as hippocampus and prefrontal cortex; restoring normal gamma rhythms by different mechanisms, such as magnetic (7), auditory (8), and visual stimulators (9), has proved to reduce cognitive symptoms, inflammation, and amyloid deposition in these same animal models. However, all the gamma-restoring strategies tested in AD models and human subjects have been performed within the slow gamma range, even if fast gamma range has shown to be related to working memory in humans (10), letting the possible effects of fast gamma-restoring strategies unexplored.

Currently, there are several non-invasive brain stimulation methods that could be used to induce or restore gamma oscillations. Between these methods, transcranial magnetic stimulation (TMS) is a very well-known method that uses electromagnetic induction in order to generate an electric field inside the brain; however, because of the high intensity of the currently used magnetic fields (near 1 T), it is unsafe to apply high-intensity magnetic pulses at fast gamma frequencies (11). Recent research has shown that very small magnetic fields (0.001 T) can also alter the physiology of the brain (12); therefore, we decided to employ the application of low-intensity magnetic fields, which have shown to be able to produce neural entrainment at the magnetic pulse frequency (13).

Here, we hypothesized that the application of small magnetic fields (5 gauss) at fast gamma frequencies over the left prefrontal dorsolateral cortex, an important node for working memory (10, 14) that has been targeted with positive outcomes in previous MCI and AD studies (15, 16), would be a safe and well-tolerated intervention that could potentially improve cognitive scores in subjects with MCI and mild AD.

To prove our hypothesis, we performed a pilot, double-blind, sham-controlled clinical trial, to assess safety and the possible cognitive effects of twice daily application during 6 months of fast gamma magnetic stimulation (FGMS) over the left prefrontal dorsolateral cortex.

A total of 95 subjects were screened for eligibility; 38 met eligibility criteria and were randomized between May 2019 and March 2020 (see Figure 2). A total of 20 subjects were allocated to the active FGMS group and 18 to the sham FGMS group. Four subjects (three in the active FGMS and one in the sham FGMS) were excluded from the primary and secondary analyses due to lack of at least 80% of attachment to intervention. During the 6 months of the intervention, four subjects presented adverse effects in the FGMS group, while just one subject presented adverse effects in the sham FGMS group. These five subjects were eliminated from the study once the adverse effects were reported and, thus, were not included in the secondary analysis. For the secondary analysis, six subjects from the active FGMS group and seven subjects from the sham FGMS were lost due to impossibility of performing accurate remote evaluation for cognitive scales. In the end, seven subjects from the active group and nine subjects from the sham group were included for secondary outcome analysis.

Baseline demographic and clinical characteristics such as age, sex, baseline scores, and relevant history of comorbidities were measured for the subjects included in primary analysis and are described in Table 1. To evaluate differences in these variables, Mann–Whitney U-test and chi-square test were performed depending on the type of variable. Except for the prevalence of stroke, no statistically significant differences were found.

Here, we present the results of twice daily application of FGMS over the left dorsolateral prefrontal cortex for 6 months in an elderly population with MCI and mild AD.

Regarding safety and usability, FGMS was adequately tolerated by most of the subjects. Only mild adverse effects were reported, and these effects remitted spontaneously without medical intervention. An important finding is that it is feasible to apply FGMS for a prolonged period (twice daily for 6 months) directly from the house of the patient. Non-invasive neuromodulation devices have shown promising results in treating neurodegenerative disorders (27); however, most of the current studies apply the neuromodulation intervention in a clinical setting for a limited period of time (27), which could affect real-world efficacy and compliance for patients and caregivers. As only three subjects in the active FGMS and one in the sham FGMS did not complete the prespecified 80% of sessions, current results show that long-term application of FGMS directly from home is feasible and has a high compliance rate.

In our sample, FGMS did not change the ADAS-Cog score after 6 months of daily application compared to sham stimulation. ADAS-Cog has been used for several years as the gold standard measurement of cognition in dementia populations; however, its utilization in subjects with MCI remains controversial, mainly because its ability to detect relevant changes in the initial stages of dementia has been discussed (28). Because of the small sample size, the possibility that ADAS-Cog was not sensitive enough to detect small changes in the studied population remains.

Other secondary outcomes were not modified by FGMS; cognitive measurements (FAB, MoCA, pVFT, and sVFT) did not achieve statistically significant differences between groups. We choose several different measurements of cognitive functions to cover a broad range of clinical changes; interestingly, the FAB score showed a crossover interaction in which the effect of time on the score was opposite depending on the group the subjects belonged to; unfortunately, because of the small sample size, statistical significance was not achieved; further trials with adequate sample size focusing on frontal clinical changes could be regarded in a future, especially in dementia with frontal involvement such as frontotemporal dementia.

Both groups showed an improvement in some cognitive measurements (MoCA, pVFT, and sVFT). This is a rather uncommon result: the studied populations tend to have cognitive decline with time. Several factors could have provoked the observed improvement in both sham and real FGMS group. Ito et al. (29) used data from several clinical trials to model the placebo response in AD; they found that with short time of interventions (6 months), some improvement falls within the 90% predicted cognitive response, which could explain our results. Longer periods of intervention and follow-up would be required to assess if this response pattern in both groups is maintained or not.

Several parameters of the stimulation could be modified for future trials testing this technology. We decided to use gamma frequencies at the fast range because fast gamma frequencies have shown to be relevant in diverse cognitive functions, especially in working memory (30) and memory encoding (31, 32); however, recent evidence shows that slow gamma frequencies at 40 Hz could be a more adequate stimulation frequency than faster frequencies (33). Entrainment ability of external magnetic pulses has been tested up to slow gamma range; this is mainly due to the difficulty of measuring fast gamma oscillations in a non-invasive way. More studies using FGMS and deep EEG recordings are needed to affirm that FGMS has an entertainment effect in humans.

Another parameter that could explain the lack of clinical changes is the intensity of the applied magnetic field. While the magnetic fields used in the present study are over the threshold for a biological effect (7, 34), the intensity of the magnetic field may not have been enough to assure a perdurable biological effect. Most evidence that affirms that weak magnetic fields are able to cause neural entrainment come from studies using magnetic fields of around 20-fold smaller than current TMS devices (35); in the present study, we used a magnetic field 200 times smaller than current TMS devices. This intensity was used because similar intensities of magnetic fields have shown to modify important pathological mechanisms of neurodegeneration in animal models such as an increase in plasticity (36), an increase in adult neurogenesis (37), and diminishing tau hyperphosphorylation (7). Translating animal findings into human neuromodulation trials encompasses great difficulties, and adjusting the proportional intensity of the magnetic field used in an animal brain to a human brain remains an important methodological challenge. Future trials using gamma magnetic stimulation should take this into consideration.

Lastly, the stimulation site could also be important for the lack of statistical results. Most of repetitive TMS trials in dementia population have used as stimulation site the left prefrontal dorsolateral cortex (27); the rationale of this decision is that the prefrontal cortex has been shown to be selectively affected in AD (38) and is a cortical region that is near and accessible through the skull (39). However, prefrontal cortex pathological changes in early stages of AD are heterogeneous and not as well-described as pathological changes in other areas such as atrophy in medial temporal lobe structures (40). More trials using FGMS focusing on different cerebral regions, such as the temporal lobe, could have different clinical results.

Our results should be regarded with caution as several issues can be noted that could have influenced our results. First, the trial was underpowered due to a small sample size; this was mainly due to the lack of follow-up in our secondary outcomes. Second, we did not include objective biomarkers as inclusion criteria for this trial, and thus, the possibility that our sample was composed of a heterogeneous sample of MCI and other types of mild dementia subjects remains. Third, the follow-up of the current trial was limited only to 6 months; while it is a considerable time frame, neurodegenerative disorders such as MCI and mild AD could demonstrate clinical changes that may have occurred after the last measurement of the variables remains. Also, immediate effects of FGMS were not assessed in this trial; a recent study showed that transcranial alternate current stimulation at slow gamma frequencies was able to improve episodic memory just after finishing stimulation (24). Our trial aimed to encounter long-term changes in scores, and we cannot discard the possibility that FGMS changes, acutely, cognitive traits. Lastly, AD has a higher prevalence in certain specific populations (e.g., women, people with history of depression), and the response to FGMS in these specific populations was not assessed in the current study. Future trials of FGMS should take into consideration the several limitations of our study.

In conclusion, FGMS over the left prefrontal dorsolateral cortex applied twice a day for 6 months resulted to be a viable intervention that can be applied safely directly from home without supervision of a healthcare provider in an elderly population with MCI and mild AD. However, in this small sample, FGMS did not statistically significantly change cognitive, functionality, or depression scores compared to sham stimulation.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Research and ethics board of National Institute for Medical Sciences and Nutrition Salvador Zubirán. The patients/participants provided their written informed consent to participate in this study.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

Actipulse Neuroscience provided the devices and financed this study.

The authors declare that GV and AR-P are currently working as part of the Science Department of Actipulse Neuroscience. Actipulse Neuroscience provided the devices and financed this study. Actipulse Neuroscience did not play any role in the study design; collection, analysis, and interpretation of the data; as well as writing of the report and decision to submit the article for publication. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.