The effects of bifrontal anodal transcranial direct current stimulation (tDCS) on sleepiness and vigilance in partially sleep-deprived subjects: A multidimensional study

In recent years, transcranial electrical stimulation techniques have demonstrated their ability to modulate our levels of sleepiness and vigilance. However, the outcomes dif-fer among the specific aspects considered (physiological, behavioural or subjective). This study aimed to observe the effects of bifrontal anodal transcranial direct current stimulation. Specifically, we tested the ability of this stimulation protocol to reduce sleepiness and increase vigilance in partially sleep-deprived healthy participants. Twenty-three subjects underwent a within-subject sham-controlled stimulation protocol. We compared sleepiness and vigilance levels before and after the two stimulation conditions (active versus sham) by using behavioural (reaction-time task), subjective (self-report scales) and physiological (sleep-onset latency and electroencephalogram power [ n = 20] during the Maintenance of Wakefulness Test) measures. We showed the efficacy of the active stimulation in reducing physiological sleepiness and prevent-ing vigilance drop compared with the sham stimulation. Consistently, we observed a reduction of perceived sleepiness following the active stimulation for both self-report scales. However, the stimulation effect on subjective measures was not statistically significant probably due to the underpowered sample size for these measures, and to the possible influence of motivational and environmental factors. Our findings confirm the ability of this technique to influence vigilance and sleepiness, pointing out the potential for new treatment developments based on transcranial electrical stimulation.


| INTRODUCTION
Daytime sleepiness represents one of the biggest challenges in our increasingly sleepless society. Sleepiness does not reflect a unitary phenomenon as: (a) it is influenced by multiple internal and environmental factors; and (b) it underlies a wide range of subjective and objective outcomes (Cluydts et al., 2002). Vigilance, defined as the ability to sustain attention over extended periods of time (Parasuraman & Davies, 1977), represents the tonic component of attention and is closely related to sleepiness. In accordance with the two-process model of sleep-wake regulation, sleepiness and vigilance are both dependent on the interaction between the homeostatic and circadian processes (Borbély, 1982).
In recent years, the investigation of the effects of transcranial direct current stimulation (tDCS) on sleepiness and vigilance levels has gained increasing attention (Annarumma et al., 2018). The most widely used protocols of tDCS involve the application of a low-intensity (0.1-2.0 mA) current stimulation through the skull by anode and cathode electrodes placed over target regions on the scalp (Fertonani & Miniussi, 2017). Conceptionally, anodal stimulation moves the resting membrane potential closer to the depolarization threshold and activates the underlying cortical area, whereas cathodal stimulation leads to the opposite inhibitory effect (Stagg & Nitsche, 2011).
Although the exact mechanisms of action of these techniques are still debated, their ability to affect the sleep/arousal pattern is consistently attributed to the modulation of the "top-down" corticothalamic pathway of sleep regulation (Frase et al., 2016;Krone et al., 2017). As the anterior cortical regions are the first to exhibit the distinctive electroencephalogram (EEG) activity of sleep-onset processes (i.e. slowing of EEG, intensification of frontal alpha and reduction of frontal theta; Gorgoni et al., 2019;Marzano et al., 2013;Werth et al., 1996), they represent the main target areas of transcranial electrical stimulation (tES) aimed at influencing the electrophysiological correlates of the wake-sleep transition (Frase et al., 2016).
Further, prefrontal regions play a pivotal role in the alerting network functioning (Langner & Eickhoff, 2013). Hence, tES protocols aimed at increasing or decreasing the neuronal excitability of these regions may also influence the cortical component of the vigilance control system (Dalong et al., 2020).
Several protocols have been implemented for using tES to promote sleep propensity and accelerate the sleep-onset process (D'Atri et al., 2015;D'Atri et al., 2016;D'Atri et al., 2017D'Atri et al., , 2019Kirov et al., 2009;Xie et al., 2021). A complementary line of research tried to investigate the effectiveness of tES as a countermeasure for excessive sleepiness and vigilance decrement (Brunyé et al., 2019). In pioneering studies, the application of anodal tDCS targeting specific areas of the prefrontal cortex was able to prevent the typical vigilance reduction across time-on-task  and to mitigate sleep-deprivation-induced vigilance drop more efficiently than caffeine (McIntire et al., 2014). A more recent study confirmed these results by examining the effects of active tDCS over the frontal areas on the behavioural and electrophysiological functioning of distinct vigilance components (Luna et al., 2020). Along the same vein, Cheng et al. found an attenuation of the subjective drowsiness and fatigue following sleep deprivation (Cheng et al., 2021). Studies on patients with idiopathic (Galbiati et al., 2016) or organic hypersomnia (Frase et al., 2015) also described the efficacy of similar stimulation protocols in reducing diurnal sleepiness as well as increasing attentional performance in a clinical population. However, they did not explore the electrophysiological pattern underlying the observed results.
Taken together, these results suggest that different tES protocols could be effective sleepiness countermeasures. However, their effects on the electrophysiological pattern have been poorly investigated and warrant further studies. To our knowledge, tES studies aimed to reduce sleepiness or increase vigilance that have simultaneously considered subjective, behavioural and electrophysiological measures are still lacking.
Thus, the present study aimed to explore the effects of bifrontal anodal tDCS on a sample of healthy subjects, in which elevated levels of diurnal sleepiness were experimentally induced through partial sleep deprivation before the laboratory sessions. We adopted a within-subjects sham-controlled design to determine the effects of the active stimulation by using: (1) EEG recordings during the execution of the Maintenance of Wakefulness Test (MWT); (2) Psychomotor Vigilance Task (PVT) to assess behavioural sleepiness; and (3) selfreported questionnaires to evaluate subjective drowsiness.
We hypothesized that the active stimulation could globally reduce the diurnal sleepiness experienced by the sleep-deprived participants, as reflected by prevention of early EEG signs of sleepiness, reduction of vigilance drops during the sustained attentional task, and lower self-rated sleepiness scores, differently from sham stimulation.

| Participants
Twenty-three healthy subjects (12 males and 11 females) aged between 24 and 37 years (mean age 29.73 ± 3.44 years) took part in the study. From the originally recruited sample, three subjects were excluded from the EEG power analyses due to the occurrence of technical problems (i.e. scarce EEG signal quality due to artefacts). The final sample considered for the EEG power analyses was composed of 20 subjects (11 males and nine females) aged between 26 and 37 years (mean age 30.35 ± 3.23 years).
All participants met the following inclusion criteria as assessed by a clinical interview: no excessive daytime sleepiness (total score on the Epworth Sleepiness Scale ≤ 10); medication-free; no presence or history of epilepsy; no neurological or psychiatric disorder; no intracranial metal implants; no daytime nap habits or any sleep disorders; no excessive consumption of neuroactive drugs or caffeine.
All participants provided written informed consent to the experimental procedure and could withdraw from the study at any moment.
The study was approved by the Institutional Ethics Committee of the Department of Psychology of the University of Rome Sapienza (Prot. n. 0000942) and was conducted in accordance with the Declaration of Helsinki.

| Experimental design
Participants were asked to keep regular sleep-wake schedules during the week before the experimental session, and to fill out a daily sleep log to control their compliance. All subjects underwent a partial sleep deprivation protocol at home (maximum 4 hr of sleep from 01:00 hours to 05:00 hours) during the night before the experimental day, monitored by sleep logs and actigraphic recordings (AMI, MicroMini Motionlogger, USA). Specifically, we checked the sleep logs and actigraphic data the following morning to verify the subjects' adherence to the deprivation protocol and define their inclusion/exclusion in the study, without further storing the acquired data.
The intake of any kind of neuroactive drugs, including coffee, tea and chocolate, or intense physical training was not allowed before the experimental session.
Our single-day experimental protocol consisted of two consecutive within-subjects sessions: one active condition (anodal tDCS) and one sham condition, separated by an interval of at least 2 hr. The order in which the real and sham stimulations were delivered in each session was randomized and balanced across subjects (i.e. 12 subjects started with real stimulation and 11 subjects with sham). As regard the subgroup of subjects considered for the EEG data analyses, 10 participants started with real and 10 with sham stimulation.
We used a single-blind protocol in which the participants were blinded to the stimulation type, whereas the experimenter who administered the stimulation was aware. Anyway, the experimenters were blinded to the specific condition during the scoring procedure.
Subjects came to the laboratory at 08.00 hours, and electrodes were fixed on their head in about 2 hr. Each session lasted 3 hr and included an identical timeline: (a) a pre-stimulation assessment; (b) the stimulation protocol (active or sham); (c) a post-stimulation assessment ( Figure 1). The pre-and post-stimulation assessment included subjective (self-reported questionnaires), behavioural (reaction-timed task) and objective (EEG recording) measures.

| Electrical stimulation
The tDCS equipment consisted of a battery-driven stimulator system (BrainSTIM, EMS Medical, Italy) and conductive-rubber square electrodes (25 cm 2 , 5 Â 5 cm) placed in sponges saturated with high-conductivity gel. In line with the aims of the study and earlier similar protocols (Frase et al., 2015(Frase et al., , 2016Frase et al., 2019), the anodes (positively charged electrodes) were individually applied bilaterally at frontal locations (F3 and F4 of the international 10-10 system), and the cathodes (reference electrodes) at temporooccipital positions (Y-cable split for stimulation and reference electrodes). A constant current (1.5 mA stimulator output) was delivered through a repetitive stimulation protocol: two consecutive blocks of 15 min with a 20 min inter-stimulation interval (30 s fade-in/ fade-out).
In the sham session, the stimulation setting was the same as the tDCS condition, but the current was reduced to zero after 30 s in order to maintain the same physical sensation at the beginning of the real stimulation. Participants reported no adverse effects of the stimulations and no perceived differences between the active and sham conditions, as verified by a post-experiment debriefing.
We used an open-source software package (SimNIBS 2.1; Saturnino et al., 2015) to generate computational modelling of the distribution of the electric field strength over the cortex yielded by the adopted stimulation protocol ( Figure 2).

| EEG recordings
BrainAmp MR plus system (Brain Products GmbH, Gilching, Germany) and Brain Vision Recorder (Version 1.10, Brain Products GmbH, Gilching, Germany) software were used to amplify and record the signals.
EEG signals were recorded with a sampling rate of 250 Hz (0.1-μV steps resolution). A high-pass filter with a time constant of 1 s and a 70-Hz low-pass filter were applied to raw EEG data (phase shift-free Butterworth filters).
The ground electrode was placed at Fpz (fronto-polar location), and the EEG signals were referenced online to the averaged mastoids (A1 and A2). Horizontal eye movements were detected by recording electrooculogram (EOG). The submental electromyogram (EMG) was also recorded for the offline artefacts detection and sleep scoring.
The EEG data were digitally stored for further offline analyses. 1. The first trial started at least 3 hr after the subject's wake-up time.
2. The trials were performed at intervals of at least 2 hr.
3. The room was maximally insulated from external light and the light source was positioned out of field of vision. The subject was seated comfortably, with the back and head supported by a pillow.
4. Instructions to the participant consisted of the following: "Please sit still and remain awake for as long as possible." Participants were not allowed to use extraordinary measures to stay awake, such as slapping the face or singing.
5. Naps and the use of neuroactive drugs before and during MWT were not allowed. Light breakfast is recommended at least 1 hr before the first trial, and a light lunch is recommended immediately after the second trial.
6. Trials were ended after 40 min if no sleep occurs or after unequivocal sleep onset (first appearance of a K-complex or spindle), identified by a sleep expert who continuously monitored the EEG recordings.

| Psychomotor Vigilance Task
The PVT is a well-established behavioural measure to assess sustained attention and objective levels of sleepiness, very sensitive to the effects of sleep loss and without learning effect (Reifman et al., 2018).
We used a 10-min version of PC-PVT software (Khitrov et al., 2014).
Participants were asked to click the left mouse button as quickly as possible every time a counter appears at random intervals.
The KSS is a self-report measure to assess subjective levels of state-like sleepiness. Subjects were asked to rate their sleepiness level on a nine-point rating scale, ranging from 1 ("Extremely alert") up to 9 ("Extremely sleepy").
The VAS-gv is a continuous measure of subjective vigilance, which takes into account the scores from four subscales (alert, sleepy, weary and effort) to obtain a global vigour score between 0 and 40.
Subjects indicated their current state (from "Not at all" to "Very much") by placing a mark on a 10-cm line for each scale. In accordance with the purposes of the study, we selectively considered the "sleepiness" scale (VAS-sleepiness).

| Quantitative EEG analysis
The EEG signals were offline high-pass filtered with the time constant of 0.3 s and low-pass filtered at 30 Hz. Ocular and/or muscle artefacts in the EEG recordings were rejected by off-line visual inspection of 2-s epochs.
Power spectra of the artefact-free epochs were computed by a Fast Fourier Transform (FFT) routine for the 26 scalp F I G U R E 2 SimNIBS simulation of transcranial direct current stimulation (tDCS) study set up. Computational modelling (SimNIBS) of the electric field strength generated by tDCS bifrontal montages with 5 Â 5 cm electrodes centred over F4 and F3 locations in the 0.5-29-Hz range (1-Hz bin resolution except for the 0.5-1-Hz bin) and then averaged across epochs (periodogram: 2 s).

| Statistical analysis
Data analysis was performed using the software package MATLAB 7.13 (The Math Works, MA, USA) and its statistics toolbox.
The physiological sleepiness measures were: (1) EEG power spectra (μV 2 ); and (2) sleep-onset latency (SOL, s), which was either the time interval from the "start" of the MWT trial to the end of the 40 min or to the appearance of a K-complex/spindle.   Table S1).

Sleep-onset latency
Results of mixed ANOVA on ΔSOL showed no effect of the Circadian phase in the active stimulation (F 1,21 = 3.245, p = 0.086), but a significant main effect of the Stimulation (F 1,21 = 7.961, p = 0.010;  Table 1).

| Behavioural sleepiness
Results of mixed ANOVA on ΔPVT-RT showed no effect of the Circadian phase of active stimulation (F 1,21 = 0.472, p = 0.499) factor, a significant main effect of the Stimulation (F 1,21 = 6.658, p = 0.017) factor ( Figure 4) and a significant interaction among them (F 1,21 = 13.672, p = 0.001; Table 1). Specifically, we found a decrement of PVT-RT following the active stimulation significantly different from their slight increment after sham. Further, planned post hoc tests showed a significant PVT-RT reduction compared with the sham-related increment only for the Active p.m. group (t = À4.057, p = 0.002), while no significant difference was found for subjects who received active stimulation in the morning (t = 0.868, p = 0.404).

| DISCUSSION
The present study aimed to assess the effects of bifrontal anodal tDCS on different outcomes of sleepiness and vigilance among partially sleep-deprived subjects. To this end, we adopted a multidimensional approach using distinct sleepiness measures (i.e. EEG recordings during MWT, a behavioural vigilance task, subjective sleepiness questionnaires).
For the first time, we showed the topographic EEG correlates of physiological sleepiness reduction following the application of bifrontal anodal stimulation. We also found a simultaneous increase in behavioural vigilance and no stimulation effect regarding self-reported sleepiness. Taken together, these results substantiate the current literature on the effects of frontal anodal stimulation on vigilance increment (Annarumma et al., 2018), and support the hypothesis of separate mechanisms underlying different aspects of sleepiness and vigilance (Cluydts et al., 2002).
Sleep drive evaluated by separate trials of MWT (before and after real and sham stimulation) allowed for assessing both macrostructural (the amount of time to fall asleep) and quantitative (spatiotemporal EEG patterns) aspects of objective sleepiness.
Many studies have consistently shown that sleep deprivation accelerates and deepens the EEG changes that characterize the sleep onset as a consequence of the increased homeostatic sleep pressure (Borbély et al., 1981, 1982Gorgoni et al., 2019Guerrero & Achermann, 2019).
The active stimulation significantly improves the subject's ability to maintain wakefulness (longer sleep latency), reflecting its power to In parallel, anodal tDCS provoked a net gain of cortical arousal, as reflected by the localized increase of rapid and desynchronized EEG rhythms (β1 and β2 frequency bands), and the spread decrease in slow EEG frequencies (δ and θ). Conversely, the sham-related EEG pattern showed the frontal increase of the distinctive markers of sleep propensity (i.e. δ) and somnolence (i.e. θ; Finelli et al., 2000), and the simultaneous reduction of higher frequency bands (i.e. β1 and β2).
From a topographical standpoint, the anterior EEG synchronization following sham condition is consistent with the well-known sensitivity of the frontal cortical areas to exhibit the earliest signs of the physiological need for sleep (De Gennaro et al., 2007;Finelli et al., 2000;Gorgoni et al., 2019;Vyazovskiy et al., 2011), especially after a period of sleep deprivation (Gorgoni et al., 2020). On the other hand, the presence of a complementary electrophysiological pattern after anodal tDCS reflects the ability of the active stimulation protocol to counteract the physiological process of sleep onset and boost the physiological levels of cortical arousal. Indeed, fast EEG frequencies are traditionally considered physiological indices of cortical arousal and motor/cognitive activation (Merker, 2013).
We can also observe a consistent correlation between the two aspects of objective sleepiness assessed by MWT. In particular, we found an association between the longer sleep latency following active tDCS and the stimulation-related (a) decrement of slow EEG frequencies (i.e. δ) and (b) increment of rapid EEG frequencies (i.e. β1 and β2).
In this vein, the bifrontal montage represents an undoubted advantage for several reasons. Firstly, as demonstrated in previous studies (Frase et al., 2016), concurrent stimulation to both hemispheres maximizes the effects as they are equally involved in falling asleep. Secondly, this specific type of montage (bifrontal tDCS with temporo-occipital references) may have engendered the additional stimulation of deep brain structures involved in the initiation of the typical EEG synchronization of falling asleep (e.g. thalamus and hippocampus; Magnin et al., 2010;Sarasso et al., 2014). Also, we chose the repetitive stimulation protocol in light of its enhanced effects over time (Monte-Silva et al., 2013) and with the ultimate goal of preserving the effects for the entire duration of the post-stimulation assessment (approximately 1 hr).
From a behavioural viewpoint, we selected PVT to assess sustained attention over time as it represents one of the most reliable instruments to test vigilance decrement in sleep-deprived subjects (Reifman et al., 2018). Consistently with our hypotheses, the performance on this RT task was better after anodal tDCS relative to sham.
The main limitation of the current repetitive tDCS protocol is represented by the short distance between the two stimulation protocols (about 4 hr). Indeed, although the early effects on neuronal and synaptic excitability last up to 3 hr, the later after-effects can last for over 24 hr (Monte-Silva et al., 2013). However, the long-lasting changes would seem relevant for enduring alterations of cerebral functions underlying learning and memory formation (Agboada et al., 2020), and therefore are not directly involved in the cortical state of arousal and excitability. In any case, the exact mechanisms explaining the long-lasting consequences of tES on the human frontal cortex are yet to be explored, and standardized guidelines on the amount of time to ensure the total wash-out are still absent.
Further, stimulation was delivered immediately before and after the sleepiness evaluation, and then its effects were inferred by comparing pre-and post-stimulation assessment. Future studies should perform online stimulation protocols (Yavari et al., 2018) to observe the direct impact of tDCS on neuronal excitability during the execution of the MWT and PVT. We should also consider that the absence of data storing related to actigraphy and sleep logs (used for the compliance check) prevented the possibility of exploring any potential correlation between the effects of the stimulation and the qualitative/ qualitative variables related to previous sleep.
Another major limitation of this study protocol is represented by the single-blind design (i.e. the experimenter was aware of the type of stimulation during the tDCS administration). Indeed, even though we adopted some precautions (e.g. making the initial sensations of the two stimulation types similar, as also revealed by the final debriefing interview), we cannot exclude some external bias due to the absence of a double-blind protocol.

| CONCLUSIONS
In conclusion, we confirmed that anodal tDCS over the frontal cortical