Co-authors André Aleman and Natasha Maurits are PLOS ONE Editorial Board members. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.
Conceived and designed the experiments: RK JHvdH MvW NMM. Performed the experiments: RK LEvN VGB. Analyzed the data: RK LEvN VGB. Contributed reagents/materials/analysis tools: JHvdH. Wrote the paper: RK LEvN VGB KHK MH JHvdH MvW AA NMM.
Current address: Institute for Risk Assessment Science, University of Utrecht, Utrecht, The Netherlands
The ‘complex neural pulse’TM (CNP) is a neuromodulation protocol employing weak pulsed electromagnetic fields (PEMF). A pioneering paper reported an analgesic effect in healthy humans after 30 minutes of CNP-stimulation using three nested whole head coils. We aimed to devise and validate a stimulator with a novel design entailing a multitude of small coils at known anatomical positions on a head cap, to improve applicability. The main hypothesis was that CNP delivery with this novel device would also increase heat pain thresholds. Twenty healthy volunteers were enrolled in this double-blind, sham-controlled, crossover study. Thirty minutes of PEMF (CNP) or sham was applied to the head. After one week the other treatment was given. Before and after each treatment, primary and secondary outcomes were measured. Primary outcome was heat pain threshold (HPT) measured with thermal quantitative sensory testing. Other outcomes were warmth detection threshold, and aspects of cognition, emotion and motor performance. As hypothesized heat pain threshold was significantly increased after the PEMF stimulation. All other outcomes were unaltered by the PEMF but there was a trend level reduction of cognitive performance after PEMF stimulation as measured by the digit-symbol substitution task. Results from this pilot study suggest that our device is able to stimulate the brain and to modulate its function. This is in agreement with previous studies that used similar magnetic field strengths to stimulate the brain. Specifically, pain control may be achieved with PEMF and for this analgesic effect, coil design does not appear to play a dominant role. In addition, the flexible configuration with small coils on a head cap improves clinical applicability.
Dutch Cochrane Centre
Magnetic stimulation of the brain is a safe and non-invasive way to modulate brain function. The best known method is transcranial magnetic stimulation (TMS) and uses strong (1–2 T) and short (<1 ms) pulses. In 1985 Barker
Weaker magnetic stimulation can be achieved with small or large coils, either commercially available or custom built. This technique often uses pulsed stimulation and is then referred to as pulsed electromagnetic field (PEMF) stimulation. Recently, some evidence has been found that exposure to MRI systems may change mood
Often commercial stimulation systems (e.g. NeoSync, Inc., PEMF Systems, Inc., CNP Therapeutics Inc.) use a limited number of small coils while in research the use of one or more Helmholz coils is often reported
The complex neural pulse (CNPTM)
The underlying mechanism of PEMF induced analgesia is poorly understood. There is some evidence for endogenous opioid mediation of PEMF analgesia in animals
Further, dopaminergic tone is correlated to mood and also sensitive to PEMF and TMS
We applied the CNP
This research has been approved by the Medical Ethical Committee of the University Medical Center Groningen. Informed consent was obtained from the subjects and the clinical investigation was conducted according to the principles expressed in the Declaration of Helsinki.
A personal computer (Pentium) and interface card (K8000, Velleman, Gavere, Belgium) were used as Arbitrary Waveform Generator (see
The interface card translates digital values into voltages. The amplifier in turn increases power to generate pulsed magnetic fields in nineteen small electromagnets radially attached to the head cap. Photo by S. Martens, consent to publication was obtained from the subject.
The computer ran a bash shell on Debian Linux (
To increase the low power output of the K8000 a DC coupled amplifier was built (
The electromagnets consisted of 25 mm long, 9 mm thick reed relays (Reed Relay 275–232, Radio Shack, Fort Worth, TX, USA) of which the reed switch was replaced
Nineteen of these electromagnets were radially attached to a regular EEG cap with a chin strap (SU-60 and KR, MedCaT, Erica, The Netherlands) using non-metallic nuts on the inside of the cap. Electromagnets were positioned according to the international 10/20 system for EEG electrodes (
All electrical equipment was powered through a medical isolation transformer (H01.96.00, Jansen Medicars, Maarssen, Netherlands) (see
The entire setup was tested with an International Safety Analyzer (601PRO, BIO-TEK Instruments Inc., Winooski, VT, USA) as a class I, type B device according to norm 601 of the International Electrotechnical Commission (IEC; 1988). Leak currents to earth were below 20% of the norm, patient leak currents were below 1.8% of the norm (always below 10 µA) at a current consumption of 0.2 A. The device passed all tests for a class I type B clinical device.
Maximum magnetic flux density was 1.45 mT at each electromagnet (see
The digital version of the CNP wave as published in the literature
DC shifted sine waves (min 0 V, max +3.47 V) of different frequencies were generated with a Function Generator (Model 110, Wavetek, San Diego, CA, USA) and a DC power supply and were then used as input to the amplifier.
The following were measured: voltage into the amplifier, voltage out of the amplifier and magnetic flux density (Gauss-/Teslameter, FH 54, with an axial Hall probe, HS-AGB5-4805, Magnet-Physik, Cologne, Germany) at the scalp side of one of the coils.
The coils acted as a low pass filter, limiting the frequency response of the system: while the amplifier had its 50% frequency around 90 kHz, the coils showed a 50% frequency at about 300 Hz (
Before clinical testing of the stimulator, a description of the equipment and a copy of the Insurance Certificate were filed with The Dutch Health Care Inspectorate (Inspectie voor de Gezondheidszorg) to comply with legislation.
The study conformed to national legislation on medical research and was approved by the Medical Ethical Committee of the University Medical Center Groningen, the Netherlands. In addition the study was registered in the Dutch Trial Register (Dutch Cochrane Centre, NTR1093,
Also, we conformed to the Dutch Personal Data Protection Act (“Wet Bescherming Persoonsgegevens” of 2001). Subjects gave written informed consent and received neither financial nor curricular incentives for their participation.
Twenty healthy volunteers, all native Dutch speakers, were recruited through advertisements on bulletin boards of the University of Groningen. Inclusion criteria were: 18–80 years old, subjectively healthy. Exclusion criteria were: neurological (e.g. epilepsy) history, psychiatric history, recent use (within four weeks) of prescription or non-prescription psychopharmaca, use of >10 units of coffee per day, use of >10 units of alcohol per day, presence in the body of MRI incompatible implants.
This was a single center, double-blind, sham-controlled crossover study conducted in the Netherlands. The within-subjects design was balanced for treatment order. All subjects received a sham and an active treatment at the same time of the day with one week in between. By using a random number generator on the stimulus PC, neither the subject nor the experimenter was aware of the nature of the treatment. The code was broken after all twenty subjects had been treated twice. Also, two field strengths were tested in order to investigate dose (field strength) effects on the outcome parameters: HIGH (amplitude 1.1 mT) and LOW (amplitude 0.4 mT). Half of the subjects received HIGH and half received LOW as their active treatment.
During a session the volunteers were seated behind a desk while wearing the treatment cap. Each session consisted of four blocks of fifteen minutes each. The blocks were identical in all aspects, except that during the first and last block only zeroes were sent to the DAC. During the second and third block either PEMF (LOW or HIGH field, one option per subject) or sham (all subjects) was applied through all electromagnets so that a total of 30 minutes of PEMF or sham stimulation was applied in each session. For sham too, only zeroes were sent to the DAC.
The applied field was measured afterwards with a tesla meter (FH 54, Magnet-Physik, Cologne, Germany) and a digital storage oscilloscope (DSO-101, Syscomp Electronic Design Limited, Toronto, Canada) and is presented in
Several tests were selected to sample the emotional, sensory, motor and cognitive domain. Two emotional inventories were administered before and after all other tests. The other tests were performed in four consecutive identical blocks of fifteen minutes each. For all parameters, the initial value was subtracted from the other values on a per subject per session basis. In case the measure was repeated within a block, the mean value per block was calculated.
Warmth detection threshold (WDT) and heat pain threshold (HPT) were taken as indicators of sensory and pain perception, respectively. WDT and HPT were measured with thermal quantitative sensory testing (tQST) with a computer controlled thermode (Thermotest, Somedic, Hörby, Sweden). The thermode was held in the non-dominant hand with the heatable surface at a thenar palmar position.
The thermode temperature started at 32°C and started warming up at 0.3°C/s at an unpredictable moment (
See text for details.
When the temperature of the thermode induced a pain sensation with an intensity of 7 on a scale from 0 (no pain) to 10 (severe pain) subjects pressed an 'escape button' which resulted in immediate and rapid cooling (3°C/sec) of the thermode to 32°C. The temperature at which they pressed the button is called the HPT and this variable was the primary outcome.
WDTs and HPTs were always measured
The speed of finger tapping was measured with a hand counter (‘hand tally’). Subjects were instructed to hold the counter in the dominant hand and to press the button with the thumb of the same hand as often as possible in 20 s. During each 15-minute block this was measured twice
The size of handwriting was assessed by the request to copy a text (single sentence of 24 words, 132 characters) by hand onto a blank piece of paper. This was done once in each 15-minute block, resulting in four time points per subjects per session. The total surface area of the written text (cm2) was determined and used for further analysis.
Once in each 15-minute block Digit-Symbol Substitution Test (DSST) of the Wechsler Adult Intelligence Scale (WAIS)
To assess emotional state during the experiment, the Dutch versions of the Positive and Negative Affect Schedule (PANAS)
After all measurements subjects were debriefed and asked to report any unusual sensations, moods or thoughts during the experiment. Also they were asked whether they noticed the treatment with magnetic fields.
For all outcomes the pre-treatment value was subtracted on an individual basis. The primary outcome was then tested across treatments with a one-sided paired t test on the difference scores. Significance was accepted at 0.05. For the secondary outcomes an exploratory analysis was done on the treatments using two-sided paired t tests on the difference scores. In this case, a conservative multiple comparisons correction (Bonferroni, 11 comparisons) led to significance being accepted at 0.0045.
Figures in the results section show the change after treatment relative to the first measurement in the same subject in the same session. Bars indicate means and standard error of the mean.
Group | n | female (%) | right-handed (%) | mean age (min, max, stdev) |
HIGH-sham | 5 | 80 | 60 | 29.4 (24, 44, 8.35) |
LOW-sham | 5 | 80 | 80 | 25.8 (20, 40, 8.07) |
sham-HIGH | 5 | 100 | 80 | 24.6 (23, 29, 2.51) |
sham-LOW | 5 | 100 | 100 | 24.4 (22, 28, 2.61) |
None of the volunteers reported adverse events or other complaints. At the exit interview they were invited to guess which treatment they had just received. There was no relationship between the actual treatment and the subjects' guess.
There were no statistical differences between data from the LOW and the HIGH treatment, therefore the two treatment groups were combined into one.
Both WDT and HPT increased over the experiment. The primary outcome parameter, HPT was increased more after PEMF than after sham (t(19) = 1.98, p = 0.0313, Cohen's d = 0.613), but for WDT there was no significant treatment effect (t(19) = 0.114, p = 0.455) (
Warmth detection threshold (WDT) was unaltered by PEMF (p = 0.455) but heat pain threshold (HPT) increased more after PEMF than after sham stimulation. * significantly different at p<0.05.
For the WAIS symbol to digit substitution task there was a non-significant effect of treatment (t(19) = 2.82, p = 0.0110, alpha crit = 0.0045) with lower performance after PEMF.
The two motor variables, used as indices of dopaminergic function, were both unaltered by the PEMF: finger tapping (t(19) = 0.920, p = 0.369) and handwriting (t(19) = 1.20, p = 0.245).
From the emotional variables the PANAS showed that the change in positive associations was negative i.e. subjects were less positive after the treatment. Likewise, the change in negative associations was positive indicating that subjects were more negative after the experiment. Despite these changes there was no evidence of a significant treatment effect:
We aimed to construct a novel device for cerebral PEMF stimulation and tested the hypothesis that the stimulation exerted an analgesic effect when applying a wave pattern known as CNP
As hypothesized the weak field PEMF treatment for 30 minutes increased HPT compared to sham stimulation. The effect size as indicated by Cohen's d is 'medium' to 'large'. During sham exposure HPT increased by approximately 1°C. In addition, the PEMF effect added approximately 0.7°C to the HPT. Thus the HPT increasing effects of PEMF were similar in magnitude to the habituation effect that developed over the course of the experiment. Taking into consideration that the thermode temperature increased by 0.3°C/s, treated subjects allowed the already hot thermode to warm up for an additional 2 to 3 seconds on average.
Our PEMF effects on tQST results are in agreement with a previous study
This study shows that the PEMF effect appeared to be quite specific for HPT. PEMF treatment had no effect on WDT, indicating that the ability to detect warmth, a non-noxious thermal stimulus, remained unaltered by the pulsating magnetic field. This is in agreement with the literature
The digit to symbol substitution task showed a non-significant treatment effect with worse performance after PEMF than after sham. The fact that this did not reach significance was because we did not have a hypothesis about cognition so this is a result from an explorative analysis with a conservative multiple comparisons correction. However, there is some biological plausibility because it was recently shown that cognitive control and sensory processing can both be influenced simultaneously by one intervention or manipulation
In order to gather information concerning the working mechanism of the induced analgesia, we also measured two emotional parameters and two motor parameters that are sensitive to dopaminergic tone. No treatment effects on the emotional state were found so we have no evidence that emotional changes were mediating the PEMF effects on HPT. We also found no treatment effects on the two behavioral indices of dopaminergic tone. Taken together these findings suggest that PEMF analgesia is not mediated by changes in emotion or in central dopaminergic tone.
The fact that pain tolerance was increased does not identify a single neuroanatomical structure as the mediating location because the level of pain tolerance is the end result of the total function of the anterolateral somatosensory system: nociceptors, thin fibers, dorsal horn, ventral commissure, spinothalamic tract, periaqueductal grey and reticular formation, ventromedial, mediodorsal and intrathalamic thalamus, insula and anterior cingulate cortex. The latter two are involved in the emotional aspects of pain such as tolerability and suffering. Therefore, these are plausible areas for mediation of increased pain tolerance and in fact a relatively recent study found support for the notion that brain activation in insula and anterior cingulate cortex as measured with fMRI was decreased by PEMF stimulation with the CNP
The mechanisms by which electromagnetic fields can influence biological systems are not yet fully understood. An abundance of mechanisms have been proposed and a large number of them have been confirmed experimentally (for review see e.g.
We found no differences between the effects induced by the two field strengths: apparently the intensities were equipotent. Dose-dependency of PEMF effects is generally very steep and has been described for different systems to occur below 1 mT
Concerning the penetration depth of our stimulation, it is often heard that TMS penetrates 1–2 cm, although H-coils can reach up to 6 cm
Thresholds for PEMF effects on living systems have been estimated at 500 nT
In terms of safety, our newly designed magnetic stimulator conformed to the assessment criteria of the Dutch Work Group for the Classification of Instruments in University Hospitals (Wibaz) and is a class I, type B device according to the IEC 601-1988 norm. This indicates that the device is electrically safe to be used on humans. With regard to neurological safety, epileptic seizures are the main serious adverse event that can potentially be induced by magnetic stimulation. However, the risk of inducing seizures is controllable because it is a function of frequency and field strength
A strength of this study is that we provide data from an actual measurement of the magnetic field whereas this is frequently omitted in PEMF reports. Additionally, we confirmed our prediction that HPT would increase after PEMF treatment and we measured many additional parameters. This study was sham-controlled and volunteers detected no difference so it was truly double-blind for the whole duration of the experiment. Although the treatment groups (PEMF-sham and sham-PEMF) were not fully balanced with respect to age, gender and handedness, the use of a crossover design precluded confusing group effects for treatment effects. In the paired design every subject served as their own control thus reducing the obscuring effects of intersubject variability.
As a limitation, the generalizability of this trial is limited because it was performed in a relatively small group (n = 20) consisting mostly of young women. Another limitation is that although the device permits considerable anatomical specificity of the PEMF stimulation, for this pilot we stimulated all locations simultaneously. Future studies should aim to elucidate the relative contribution of the individual electromagnets.
In summary, we built a magnetic stimulator capable of producing fluctuating magnetic fields with arbitrary temporal patterns within the 0–300 Hz frequency range. The use of an established coordinate system allows studies with anatomical specificity and integration with existing (EEG) literature. The use of nineteen small electromagnets makes it possible to stimulate specific neuroanatomical targets with the aim of modulating their function. This setup allows double-blind, sham-controlled experiments with arbitrary wave shape magnetic stimulation. These advantages, in addition to low cost and high safety, make this technology widely applicable for functional and clinical studies of the brain. As expected PEMF stimulation of the brain with this device caused increased pain tolerance in healthy subjects. At the same time, sensitivity to non-noxious thermal stimuli remained unchanged. We found no evidence for changes in emotional state and motor parameters that correlate with dopaminergic tone, thus it is unlikely that these would have mediated the changes in pain sensation.
(TIF)
We are grateful to K. Vaartjes for design and construction of the amplifier, to Y. Bloemhof for advice on construction and safety testing, and to L. Nanetti for programming support, as well as to the subjects, who all participated pro bono in this research, and to T. Nijboer for comments on the manuscript.