Percutaneous neurostimulation (PNS) of the spinal cord has been developed in the last decade for the management of intractable pain [1,2], but also for the treatment of several neurological disorders such as spasticity [3], parkinsonian tremor [4] or epilepsy [5], More recently, PNS has been explored for the treatment of intractable headaches [6-11].

Transcutaneous neurostimulation (TNS) is a classical technique which has demonstrated its efficacy in the treatment of pain [12,13] and is nowadays largely in use in pain clinics and physical therapy centres. It has the advantage of being non-invasive, safe and almost devoid of adverse effects contrary to PNS which needs a surgical intervention to implant the electrodes and the neurostimulator.

TNS at cephalic sites has been technically difficult and usually rather painful. STX-Med company has recently developed a headset for TNS of supratrochlear and supraorbital nerves, both branches of the ophthalmic division of the trigeminal nerve (V1), making the technique comfortable and easy to use [14]. Consequently, the utility of TNS in the treatment and prevention of headaches and migraine has been investigated [15] and several clinical trials are underway. Subjects enrolled in those trials have repeatedly reported that supraorbital TNS tended to affect vigilance and decrease attention with a tendency to fall asleep during the stimulation.

Cephalic electrical stimulation has been used many years ago to induce sleep or decrease anxiety. The method known as "Cranial Electrotherapy Stimulation (CES)", also called transcranial or transcerebral electrostimulation differs from TNS in that its objective is to generate different types of electrical currents through the head and not to specifically stimulate cranial nerves like TNS. For this purpose, CES uses generally an anterior frontal or a jaw electrode and a posterior electrode placed over the mastoid process [16,17]. CES was reported to have some effects on anxiety, depression and insomnia [18-20].

The Psychomotor Vigilance Task measures the reaction time (RT) and is considered as the gold standard for measuring sleepiness [21]. That it is readily reproducible is demonstrated by the fact that during the blank condition (BC) the change compared to baseline was less than 1.5%. Sham (SC) and LFN induced non significant increases in RT of respectively 8.9 ms and 8.6 ms. By contrast, HFN increased RT by an average of 36.7 ms, i.e. by more than 10%. Critical flicker fusion frequency is known to decrease with fatigue. While unchanged during SC and minimally increased during BC (+ 0.9 Hz), it increased during LFN (+ 1.9 Hz) possibly suggesting a mild increase in vigilance. Again HFN contrasted with all other conditions by a marked decrease (-4.6 Hz) in CFFF, indicating a decrease in arousal. This result is concordant with that of the subjective fatigue rating on the Fatigue Visual Numerical Scale (FVNS). The subjects rated their fatigue higher during all experimental conditions than at baseline, which was not significant and might be related to the mental strain due to the recordings or to a learning effect in using the numerical scale. However, the increase of the FVNS score during HFN was three times greater (+ 72.1%). The d2 test for attention and concentration was in our study the only one for which the HFN condition induced no significant effect. Nevertheless the numerical changes during HFN are in line with the other results as they show a lower number of total letters marked and of correct letters marked as well as a higher number of errors. The lack of significance could have at least two explanations. First, the d2 test was administered at an earlier time point (between 10 and 15 minutes) during the experimental condition compared to the other tests (from 15 minutes onwards). The duration of HFN might thus not have been long enough to produce significant d2 test changes. Second, this test was performed only once to avoid a learning effect and the pre- and per-condition comparison had therefore to be replaced by a comparison between conditions, hence weakening the sensitivity of the test to detect a change.

To the best of our knowledge, this is the first time that the effect of transcutaneous neurostimulation on arousal and fatigue was studied in humans and there are no similar studies available in animals. The neurobiological mechanisms through which HFN induces sedation remain therefore speculative. Some insight can nonetheless be gained from the studies of transcutaneous neurostimulation in Alzheimer's patients and from those in experimental animals of the central nervous system consequences of electroacupuncture. A Dutch group reported in a series of publications that transcutaneous electrostimulation was able to improve memory, alertness [26,27] and rest-activity rhythm [28] in Alzheimer's disease. This effect was attributed to activation of the hippocampus and the suprachiasmatic nucleus both by direct spinal cord afferents [29] and via the dorsal raphe nucleus and locus coeruleus [30,31]. Although vigilance was not specifically measured in these studies, the observed cognitive and behavioural effects would suggest increased arousal and vigilance rather than sedation like in our study. This opposite effects can probably be explained by the different stimulation protocols. First, Alzheimer patients received transcutaneous neurostimulation over paravertebral back muscles daily during 6 [26] or 3 hours [27,28] for 6 weeks while we used a single 20-minute session of supraorbital neurostimulation. In a more recent randomized sham-controlled pilot trial of right median nerve stimulation, Scherder et al [32] found no significant effect on memory in Alzheimer's disease and the same group reported that cranial electrostimulation had no effect on rest-activity rhythm neither at low frequency [33] nor at high frequency [34]. More interestingly, we found a hypnotic effect with high frequency (120 Hz) stimulation, whereas the beneficial effects in Alzheimer's disease were obtained with burst of stimuli (9 pulses at 160 Hz) delivered at a low frequency of 2 Hz, a frequency that in our study concordantly increased critical flicker fusion frequency. One may assume that high and low frequency stimulations can have different effects on central nervous system structures and thus on arousal, but this remains to be proven in an adequate study.

Transcranial direct current stimulation (tDCS) is able to modulate cortical activity under certain conditions and in certain brain areas. It is extremely unlikely, however, that the supraorbital TNS used in this study influences directly the underlying brain structures, i.e. the frontal lobes, for at least two reasons. First, The small electrode surface (7 cm²) and distance between the two electrodes (5 mm) restrict the skin surface affected by the current as well as current penetration into deeper structures. Second, the TNS applied current is composed of biphasic rectangular impulses with an electrical mean equal to zero, while tDCS uses a direct current. The current characteristics and the mechanisms of action are thus different between trigeminal TNS and tDCS. Moreover, in a recent study [35], weak transcranial electrical DC or AC currents over the prefrontal cortex had no effect on mood or EEG in healthy subjects. Interestingly, sleepiness was reported rarely both in the active (0.11%) and sham stimulation groups (0.08%).