Transcranial magnetic stimulation cortical oscillations and improve cognition in obstructive sleep apnea patients

Abstract Background Transcranial magnetic stimulation (TMS) is a noninvasive tool to improve cognition. Relevant clinical studies are mainly focused on neurological and psychiatric diseases. However, cognition decline and psychiatric disorders are popular in obstructive sleep apnea (OSA) patients. We aimed to investigate the effect of TMS over the left dorsolateral prefrontal cortex (DLPFC) on cognition test performance and to compare the changes in quantitative electroencephalogram (EEG) before and after stimulation for OSA. Methods This study recruited 42 OSA patients diagnosed with polysomnography according to American Academy of Sleep Medicine guidelines. TMS (intermittent theta‐burst stimulation paradigm; 2 s on, 8 s off, 600 pulses*3, intermittent 15 min) was performed on the DLPFC. Cambridge Automated Neuropsychological Test Battery was used to assess cognitive performance. EEG oscillations were computed via power spectral density with MATLAB software. Results Real‐TMS group displayed a significant improvement in visual memory, sustain attention performance, as well as the outcome of working memory. However, the executive function of latency was changed in both groups. Furthermore, TMS resulted in a significant increase in the relative power spectral density of the theta band and beta band in the parietal, temporal, and anterior regions, respectively. Conclusions In summary, our findings indicate that TMS can safely modulate cortical oscillations and improve cognition in OSA patients. In the future, TMS can be utilized as an alternative treatment option to improve cognition in OSA patients.


INTRODUCTION
Obstructive sleep apnea (OSA) is a sleep disorder and a highly prevalent disease. Approximately 26% of adult populations suffer from sleep troubles with OSA (Patil et al., 2019). It is characterized by obvious restriction of airflow (hypopnea and apnea) or episodes of cessation during sleep, which is often associated with detectable arousals on electroencephalography and decreased blood oxygen saturation.
OSA is a systemic disease accompanied by many comorbidities. It is associated with all aspects of metabolic syndromes, dyslipidemia, hypertension and obesity, myocardial infarction, congestive heart failure, and stroke (Jordan et al., 2014). All these comorbidities, together with OSA, impair cognitive function, including working memory, executive functions, and attention (Legault et al., 2021). This not only affects the quality of life for patients, but also associates with increased risks of workplace accidents and traffic crashes. As reported, OSA patients have an increased risk of car accidents by 2-10 times (D'Rozario et al., 2013). Numerous experiments have established that OSA is a high-risk factor for Alzheimer's disease (AD) (Jorge et al., 2019). Continuous positive airway pressure (CPAP) is the gold standard treatment of OSA, and is shown to ameliorate the severity of OSA, while the reversibility of CPAP on the above cognitive impairments is controversial (Bubu et al., 2020;D'Rozario et al., 2022;Patil et al., 2019). Therefore, it is a critical topic to find new methods to improve the cognition of OSA patients.
Transcranial magnetic stimulation (TMS) is a noninvasively interference that is utilized to directly study the physiology and activity of the intact human brain (Nevler & Ash, 2015). TMS modulates the brain excitability synaptic plasticity in the cortical by producing electromagnetic microcurrents. The effects not only focus on subcortical areas to the stimulation site, but also spread into remote and functionally connected corticals (Rabey et al., 2012). TMS has been used to improve cognition in various psychiatric and neurological diseases, including AD (Eliasova et al., 2014;Rabey et al., 2012;Rajji, 2019). The factors affecting the efficacy include stimulus frequency, intensity, stimuli total number, and the interval between stimuli. According to the stimulation frequency, the TMS mode is divided into low frequency (< 1 Hz) and high frequency (> 1 Hz). Intermittent theta-burst stimulation (iTBS) pattern is a pattern of high-frequency TMS that prolongs the after-effects of the cortical changes despite its shorter duration lower stimulus intensity (Vallence et al., 2015). The left dorsolateral prefrontal cortex (DLPFC) is a pivotal hub for network integrations and participates in memory, attention, and executive functions (Xia et al., 2021). DLPFC is one of the most common target stimulation sites in clinical and scientific research.
Cortical oscillations reflect the synchronization of neuronal populations in different frequency bands. The brain function is based on the fundamentals of various frequency bands in oscillation (Henry et al., 2014). Topographic electroencephalogram (EEG) is practicable for implementation in clinical settings as it can assess physiological changes of oscillatory activity in a sensitive and noninvasive way.
Moreover, EEG provides time-sensitive recordings of neural activity.
Quantitative analysis of the electroencephalogram (qEEG) has been reported to be effective for identifying small changes that contribute to functional brain tissue alterations (Corsi-Cabrera et al., 2006). The power spectral analysis is a method of qEEG. The application of qEEG in OSA research has become increasingly popular in recent years. While most of the research has focused on the EEG in sleep, EEG slowing is widely observed (D'Rozario et al., 2013;Liu et al., 2021). By combining EEG and TMS, we can further understand the physiological mechanism underlying the influence of TMS on the excitability of the oscillatory activity. There has previously been no study that investigated the changes in qEEG and cognition performance after TMS stimulation in OSA patients.
In summary, previous studies have established that TMS can produce physiological changes in the brain and the underlying mechanism is related to modulating brain oscillations of different frequency bands.
In this pilot study, we aimed to preliminary investigate whether TMS could modulate the cognitive function and alter the oscillatory activity of OSA patients.

Anthropometric and sociodemographic variables
Variables including date of birth, age, sex, education level, body mass index (BMI), history of marriage, and history of smoking and alcohol consumption were collected. PSG data included minimum SaO2%, F I G U R E 1 Channel locations. 32 Ag/AgCl electrodes placed according to the 10/20 system and a cephalic (Fpz) location as the ground, referenced to left and right mastoid.
Oxygen Desaturation Index (ODI), sleep structure (the percentage of sleep time in each stage), and Apnea-Hypopnea Index (AHI).

Experimental design
This study utilized a single-blind, randomized, sham-controlled design ( Figure 1). Participants were randomly assigned to the TMS group (real-TMS) and control group (sham-TMS) before experimenting. Subjects were seated in a comfortable armchair in a quiet, dimly light, and sound-proof room with their body in a relaxed position, and the rest-

Rapid visual information processing
Subjects were asked to find target number sequences (i.e., 2-4-6, 3-5-7, 4-6-8) among a set of pseudorandom numbers, and press the key to record. The results included: total hits (RVPTH), the number of target sequences correctly responded to; probability of hit (RVPPH), calculated from hits/(hits +misses); and RVPA, the signal detection measure of sensitivity to the target, regardless of response tendency. Probabilities and sensitivity were calculated using the Signal Detection Theory measures related to sustained attention.

Spatial working memory
Subjects clicked on the squares on the screen to "open" them, find the blue signs, then fill the holes on the right side of the screen with the blue signs. Subjects did not click the same square twice in the same round of searching for the blue mark. As the number of squares increases, the difficulty of this task gradually increases. The outcome measures for the SWM test include: total error (SWMTE), the number of times a box is selected that is certain not to contain a blue token and, therefore, should not have been visited; strategy score (SWMSS), obtained by counting the number of times the subject begins a new search with a different box. A low strategy score indicates good performance, suggesting that there are few repetitive clicks on the squares that have been found in blue.

Pattern recognition memory
The screen displayed a series of visual patterns, one at a time. Then, the participant was asked to choose the pattern that just appeared from a novel pattern. The results for the PRM test included: mean latency to correct (PRMMLC), reflecting the speed of reaction; correct percentage (PRMCP), the number of correct responses.

One touch stockings
Two patterns were displayed on the screen, and each pattern showed three balls of different colors. The participants had to plan in their minds how many moves the three colored balls would need to be moved to align the positions of the colored balls on the upper and lower monitors. The outcome of OTS included: mean latency to correct (OTSMLC), measured from the appearance of the balls on the screen until the correct box was touched; mean choice to correct (OTSCC), the mean number of choices that the subject made on each problem to make the correct choice.

TMS
Participants included in this study received TMS, with an 80% rMT to the DLPFC. The rMT was defined as the lowest intensity and was estimated before TMS. The mode of TMS was iTBS consisted of 200 busts.
The frequency of each bust was 5 Hz applied every 200 ms for 2 s and F I G U R E 2 Overview of the experiment design. Of the 46 participants, three subjects did not fulfill the inclusion criteria. The 43 subjects left, were randomly assigned to the real-TMS group (n = 22) and the sham-TMS group (n = 21). In the sham-TMS group, one was dropped out due to the inability to follow the protocol.
repeated every 10 s for a total duration of 191 s, each burst consisting of three stimuli pulses delivered at a frequency of 50 Hz, a total of 600 stimuli per session. To achieve a better clinical after effect, three typical iTBS were delivered three times at 15 min with a total of 1800 stimuli (Nettekoven et al., 2014). The sham group used a dedicated placebo coil, which was the same shape as the real coil and produced the same stimulating sound and sensation.

EEG recording and analysis
Baseline EEG was collected 10 min per-TMS treatment, and post-TMS treatment was collected within 10 min after treatment, each for 10 min. Throughout the monitoring process, the subjects were monitored with electroencephalographic signs of sleepiness tendency.  then rereferenced to the right and left mastoids. MATLAB was also used for further EEG spectrum analysis.
Neural oscillations were quantified based on fast-Fourier transforms using a Hanning window. Then, the comment power spectrum was found on each lead, and the mean relative power density was computed for each of the following frequency bands: beta (14−29.9 Hz), alpha (8−13.9 Hz), theta (4−7.9 Hz), and delta (

Statistical analysis
Statistical calculations were performed using SPSS 22.0. For each variable, the Gaussian distribution was evaluated by the Shapiro−Wilk test. Data were expressed as mean ± standard deviation (SD) or median with 25th and 75th percentile.
Distributions of baseline demographic and neuropsychological variables were compared using the independent sample t-test or twosample nonparametric Wilcoxon rank-sum test (Table 1). To objectively quantify and evaluate the short-term effects of rTMS on cognition performance and qEEG measures, we compared clinical changes per-and post-stimulation of the two groups with paired sample test. All tests were performed as two-tailed tests.
The association between the clinical features and qEEG measures or cognitive outcomes was determined by exploratory analyses using Pearson's correlation coefficients.

Demographic and baseline data of the subjects
Out of the 46 participants who were screened, three subjects did not fulfill the inclusion criteria. Among the remaining 43 subjects, 22 were randomly assigned to the real-TMS group and 21 to the sham-TMS group. In the sham-TMS group, one dropped out due to incompliance to follow the protocol. No significant differences were found between the two groups in terms of demographics (age, education level, BMI), PSG data (AHI, ODI, minimum SaO2%), neuropsychological battery (MoCA, MMSE), Epworth sleepiness scale, PSQI, or rMT (p > .05) ( Table 1).

Effects of iTBS on cognitive performance
The scores of CANTAB tests of the two groups at baseline were comparable. No significant difference was observed between the two groups (p > .05) ( Table 2).
Paired-sample comparisons of the change in CANTAB performance were conducted in two groups ( Abbreviations: OTS, one touch stockings; PRM, pattern recognition memory; RVP, rapid visual information processing; SWM, spatial working memory.  (Figure 6).

Effects of iTBS on power spectral density
EEG data from two of the 42 participants were excluded due to excessive EEG artifacts. The difference (p < .05) in power spectral density of brain region between real-TMS of per-stimulation and post-stimulation The change of RVP outcome before and after TMS in two groups. There was a significant improvement in the real-TMS group for RVPA (a), total hits (b), probability of hit score (c) after iTBS, but not in the sham group. * p < .05; ** p < .01; n.s means p > .05.

F I G U R E 4
The change of SWM outcome before and after TMS in two groups. The change of the strategy score (a), and the time to first response (b) both groups. * p < .05; n.s means p > .05.

F I G U R E 5
The change of PRM outcome before and after TMS in two groups. Only the real-group participants showed an improvement in the correct latency (a) and percent correct score (b). * p < .05; ** p < .01; n.s means p > .05.

F I G U R E 6
The change of OTS outcome before and after TMS in two groups. In both groups, changes in the latency to correct are statistically significant (a), while neither group's performance improved on the choice to correct (b). *** p < .001; n.s means p > .05.

Real-TMS (n = 21) Sham-TMS (n = 19)
Brain is presented in Table 4. There was no significant difference in relative power spectral density in the two groups of participants at baseline.
The change after TMS in the real-TMS group was observed in the theta band and beta band. Alterations in relative theta band power were in temporal and parietal areas. However, the change in the beta band was limited to the anterior regions.
Contrary to this, no statistically significant change was found in the sham group after TMS in any of the four bands.

DISCUSSION
OSA is frequently accompanied by neurocognitive impairment, often involving attention, memory, and executive function (Canessa et al., 2018;Olaithe et al., 2018). The prevalence of mild cognitive impairment (MCI) in OSA ranges from 11% to 17%, and the severity of cognitive impairment is not necessarily equivalent to the severity of OSA (Liguori et al., 2020). CPAP treatment can decrease hypoxia degree and improve daytime sleepiness, and is the gold standard treatment of OSA, while the impact of CPAP on cognitive enhancement is modest and controversial (Bubu et al., 2020). Saunamaki et al. (2010) found that after 6 months of CPAP treatment, OSA patients' visuospatial organizational skills failed to improve. Furthermore, a meta-analysis was performed to assess the efficacy of CPAP in improving cognitive function in OSA, but failed to demonstrate any significant differences between CPAP and control groups in all domains of cognition function (Patil et al., 2019).
EEG is a promising tool for the early detection of cognitive impairment, while the study conducted by D' Rozario et al. (2022) found that the EEG slowing ratio increased in patients after CPAP treatment, suggesting the limitations of CPAP on cognitive improvement and brain damage. Substantial heterogeneity in outcomes and lower adherence to CPAP treatment are additional challenging issues.  , 2011). Theta band is a core electrophysiological mechanism in internally directed attention, learning, and memory, and is also essential to executing conflict network (Herweg et al., 2020). In our study, variations in the power spectrum were found in theta and beta bands, increased post-stimulation compared to pre-stimulation. The relative power spectrum of the alpha band in the anterior, central, parietal, and temporal regions increased after stimulation, but failed to reach statistical significance. These alterations in EEG oscillations may contribute to enhanced cognitive performance, but more research is needed to investigate further.
These results suggest that the iTBS over DLPDC potentially modulates cognition and modifies the oscillatory activity of OSA patients, providing the framework for additional in-depth investigation.

Tolerability and safety
iTBS at 80% of the rMT was safe and well-tolerated. Side effects were mild and transiently prevailing (Lefaucheur et al., 2020). Theoretically, TMS may induce seizures, but the risk is very low. Patients with a history or a family history of epilepsy were excluded from our study. Only two in the real-TMS group and one in the sham-TMS group complained of a slight head discomfort which was relieved after half an hour of rest. There were no serious adverse events in either treatment group.
Therefore, applying TMS intervention to OSA is safe.

Strengths and limitations
First of all, this is the first randomized controlled study to explore the effect of TMS on OSA patients. Second, we used CANTAB for cognitive assessment, which is an objective and sensitive evaluation measure.
However, these neurological assessments are not widely used. In addition, we did not include female subjects and did not group subjects by disease severity. Meanwhile, due to the lack of MRI data, we cannot more intuitively assess the impact of TMS on brain structure. Finally, studies with larger sample sizes are needed to further determine the safety and efficacy of TMS in the long run. At the same time, the effects of different stimulation modes on different cognitive regions also need to be further explored.

CONCLUSIONS
Our results demonstrate that TMS can influence cerebral cortical activity and improve cognitive performance. TMS is a promising noninvasive therapy for cognitive enhancement in individuals with OSA.

CONFLICT OF INTEREST STATEMENT
All of the authors declare that they have no conflict of interest.

DATA AVAILABILITY STATEMENT
Data could be obtained upon reasonable request to the corresponding author.