!
!
!
!
!
Shining!light!on!the!
neurophysiological!mechanisms!!
underlying!hyperarousal!in!insomnia!!
!
A!combined!TMS!and!MRS!study!
!
!
!
!
Katharina!Müller!
UvA!Student!no.:!10867562!
!
!
Supervisor:!Rick!Wassing!
CoIassessor:!Ilja!Sligte
!
! ! ! ! ! ! ! !ABSTRACT
Up to one in ten of the general population suffers from insomnia, a sleep disorder characterized by problems with the initiation and maintenance of sleep and an overall reduced sleep quality. Despite the sleep deprivation, insomnia patients still suffer from long sleep latencies. This seeming paradox has been ascribed to a constant state of hyperarousal. Cortical excitability and intracortical facilitation, as measured by transcranial magnetic stimulation (TMS), have indeed been shown to differ in insomnia patients. Moreover, compared to controls, levels of the inhibitory neurotransmitter GABA differ in insomnia, and relate to hyperarousal. MRS-assessed GABA concentrations have been related to TMS measures of cortical excitability in the motor cortex.
Since the underlying neurological mechanism of insomnia are still unclear, the aim of this study is to specifically relate TMS-assessed cortical excitability patterns with GABA levels in insomnia. We present a study on 13 patients suffering from moderate to severe insomnia, and 40 controls (age: 48 ± 15 years).
Paired TMS with inter-pulse intervals (IPI) of 2 ms and 4 ms are shown to elicit a different MEP amplitude compared to paired stimulations with 1 ms inter-pulse interval. This difference might be due to effects of refractory period in 1 ms IPI pulses. And since there is a trend towards a significant difference between insomniacs and controls in relative SICI amplitudes only if pulses at 1 ms IPI are included, we cautiously suggest that refractory period mechanisms might differ in insomnia. In accordance with a previous study, there is trend towards a negative correlation between MEP amplitudes at IPIs of 1 ms and GABA levels in the handknob. GABA levels however do not differ between insomnia patients and controls. Higher concentrations of GABA in the thalamus are associated with higher resting motor threshold. This higher motor threshold correlates with age. However, GABA levels in the motor cortex do not correlate with the resting motor threshold. This suggests that not GABA levels but other neurological mechanisms lead to reduced cortical excitability in elderly people. The influence of thalamic GABA might have an influence on this process in the motor cortex.
In conclusion, this study shows correlations between TMS measures of cortical excitability and MRS measures of GABA, and further evidence for a different underlying neurophysiological mechanism for paired-pulse TMS at 1 ms compared to longer intervals. Underlying neurological mechanisms involved in insomniac hyperarousal might be a disturbance in refractory period mechanisms in the motor system. It is highly recommended to collect further evidence with higher sample sizes for this hypothesis.
INTRODUCTION
Insomnia and hyperarousal
They toss and turn in bed, sometimes for hours. And when sleep has finally come, it does not take long until they wake up again. Problems with the initiation and maintenance of sleep and an overall reduced sleep quality are the most prominent symptoms of insomnia. Up to one in ten people of the general population suffer from this sleep disorder; numbers differ, depending on the population and the criteria used to diagnose insomnia (Morin et al., 2006; Ohayon, 2002). The underlying neurological mechanisms of this highly prevalent disorder are however not clear.
Patients with primary insomnia report increased fatigue, but contrary to people with sleep deprivation, they still suffer from long sleep latencies (Bonnet & Arand, 1996). These seemingly paradoxical symptoms have been attributed to an overall state of physiological hyperarousal in insomnia (Riemann et al., 2010; Bonnet & Arand, 2010). This hyperarousal manifests itself in elevated body temperature and elevated whole body metabolic rates not only during sleep, but also throughout the day; and insomniac sleep is characterized by high-frequency EEG-activity and an increase in heart rate (Bonnet & Arand, 1997; Bonnet & Arand, 2010). An animal model showed that the sleep of rats with insomnia-like symptoms is characterized by the simultaneous activation of the sleep-promoting areas in the forebrain, the arousal system (HPA-axis) and the limbic system. Lesions in the arousal and limbic system of the insomniac rats restored both REM- and non-REM-sleep (Cano et al., 2008). However, improving the sleep pattern in insomnia patients via sleep therapy does not decrease high cortical excitability (van der Werf et al., 2010). Taken together, these findings suggest that hyperarousal – instead of merely low sleep quality – is associated with the insomnia-specific symptoms and is a necessary factor to modify in order to restore sleep in insomnia.
Insomnia is usually treated by different forms of cognitive behavioural therapy (CBT; Morin et al., 1999), or by non-benzodiazepine hypnotic drugs such as zopiclone, which increase the transmission of Gamma-Aminobutyric acid (GABA; Chaplin et al., 2013), with CBT reported to be more efficient in short-term and long-term measures than zopiclone (Sivertsen et al., 2006). This latter finding arises the question if current pharmacological treatments for primary insomnia actually tackle the sleep disorder by its root, i.e. hyperarousal. In fact, the underlying neurophysiological mechanisms of hyperarousal in insomnia are still unclear.
Transcranial magnetic stimulation
Transcranial magnetic stimulation (TMS) offers the possibility to quantify excitability of the cerebral cortex in vivo, especially of the motor cortex and the corticospinal tract. TMS delivers short-lived magnetic pulses to the skull, which induce local electric currents, causing a stimulation of the neurons in the targeted brain area. Following van der Werf’s study (2010), we applied TMS over the part of the primary motor cortex, which enervates the contralateral hand muscle, the so-called handknob. The motor evoked potential (MEP) is a direct measurement of the strength of the muscle response elicited by a single TMS-pulse and its time lag. The amplitude indirectly refers to the excitation state of output cells in the motor cortex, with larger amplitudes reflecting increased cortical excitability, and possibly also increased glutamatergic activity (Di Lazzaro et al. 2003).
More advanced is the application of two successive pulses, which probes the intracortical control over excitability. The first pulse is below, and the second pulse above the threshold to elicit a MEP – though variations of this protocol have been applied (e.g. Salas et al., 2014). With this design, the first sub-threshold pulse (S1) modulates the response of the second supra-threshold pulse (S2). Whether the effect is of inhibitory or excitatory nature, is defined by the inter-pulse-interval (IPI). Short-interval intracortical inhibition (SICI) is elicited by
IPIs of 1-5 ms, intracortical facilitation (ICF) is associated with moderate latencies (7-20 ms) and long-interval intracortical inhibition (LICI) takes place at IPIs in between 50 to 200 ms.
TMS and insomnia
Despite equal motor thresholds between insomniacs and controls, van der Werf et al. (2010) found higher cortical excitability in insomniacs, expressed by higher responses to supra-threshold single-pulse stimulation at 120 % MT. In the paired-pulse condition, insomnia patients exhibited significantly higher ICF than controls. They did not differ in SICI. Expressing the paired-pulse responses not in absolute terms, but with respect to the single pulse at 120 % MT, this effect reversed and remained significant only for ICF at IPIs of 11 and 15 ms. Increased ICF thus appears to be ascribed mainly to an enhanced motor response to the supra-threshold pulse at 120 % MT. These findings are confirmed by a recent study by Salas et al. (2014), showing an increase only in ICF but not in LICI or SICI. However, this study differed from van der Werf’s experimental design in that for SICI and ICF, the second, supra-threshold pulse was set such that it elicited ~1 mV MEPs.
Magnetic resonance spectroscopy
The present study was set out to replicate the findings of van der Werf et al. and relate them to the abundance of GABA, as measured by magnetic resonance spectroscopy (MRS). MRS relies on the same principle as magnetic resonance imaging (MRI). The frequency of the magnetic resonance signal provides information about the concentration of chemicals in the tissue.
MRS and insomnia
GABA acts as an inhibitor on the postsynaptic neuron, and has been suggested to be heavily involved in promoting and maintaining sleep, by acting on thalamocortical pathways and
suppressing the arousal systems (Wafford & Ebert, 2006). There is evidence for a decrease in global GABA levels – in areas including the thalamus – in insomnia (Winkelman et al., 2008) and in hyperarousal in general (Centonze, 2005), but also for an increase of GABA in insomnia (Waffors & Ebert, 2006). Since GABAa-receptors are highly expressed in the thalamus (Bateson, 2006) and the thalamus is involved in suppressing the arousal system to promote sleep (Sforza et al., 1995), we measured GABA-levels in the thalamus.
TMS and MRS
GABA is involved in intracortical inhibition mechanisms. SICI is likely driven by GABAa-receptors (Ilić et al., 2002), but it is also influenced by GABAb-GABAa-receptors, as is LICI (McDonnell et al., 2006). ICF is driven by intracortical glutamatergic excitatory circuits (Liepert et al., 1997; Ziemann et al., 1998; Schwenkreis, 1999). GABA also seems to be involved in ICF; it has been shown to suppress ICF (Ziemann et al, 1996a). There is evidence that global cortical excitability as measured by TMS is related to MRS-assessed glutamate-levels in the primary motor cortex (M1) (Stagg et al., 2011).
Due to the role of GABA in ICI and ICF, we assess GABA-levels also in the handknob. It has to be noted however that it is not very clear how local GABA-ergic synaptic activity relates to the total concentration of GABA measured by MRS in a relatively large volume of cortical tissue, and in which relationship this stands to cortical excitability measured by TMS (Stagg et al., 2011). There is evidence however that the 1 ms IPI reflects extracellular GABA, which can be measured by MRS (Stagg et al., 2011; Ziemann et al., 1996b). It has been suggested that pulses at IPIs of 1 ms do not elicit ‘true inhibition’, but the response is rather a result of the refractory period of axons excited by S1 (Peurala et al., 2008). Only at IPIs ranging from 1.5 ms (Ilić et al., 2002) or 2 ms (Ziemann et al., 1996a) to 4.5 ms (Peurala et al., 2008), synaptic inhibition is at play.
To our best knowledge, MRS-studies have not found a difference in glutamate levels in insomniacs, only of GABA (e.g. Winkelman et al., 2008). This seems counter-intuitive, since glutamate has been shown to play a role in ICF (Liepert et al., 1997), and van der Werf et al. (2010) and Salas et al. (2014) reported differences between insomniacs and good sleepers only in ICF, but none in SICI or LICI. It is possible that the currently applied means for acquiring and analyzing MRS-data are not sensitive enough to detect glutamate. At 1.5 Tesla, GABA and glutamate MRS-signals result in a complex of peaks, making it impossible to separate the two (Soares 2009). In this study however it was scanned at 3 Tesla, and there are no studies to our knowledge indicating if this issue persists at 3 Tesla. Another possible explanation for these null results is that glutamate might not influence ICF in insomniacs. Another candidate is N-Methyl-D-aspartate (NMDA), which mimics glutamate, but only has affinity with NMDA-receptors: An NMDA-receptor antagonist has been found to reduce ICF (Schwenkreis, 1999). The concentration of NMDA however cannot be assessed with MRS.
Despite the suggested role of GABA in hyperarousal and insomnia and the relationship between GABA- and TMS-assessed cortical excitability, no study has investigated this relationship in insomnia. We here report a study of 13 patients classified as suffering from moderate to severe insomnia according to the Insomnia Severity Index (ISI; Morin & Barlow, 1993) and 40 matched controls. Our objective was to investigate possible underlying neurological mechanisms of hyperarousal in insomnia. We hypothesize that cortical excitability is associated with GABAergic levels in M1. Moreover, we hypothesize that hyperarousal in insomnia can be explained by an increase in cortical excitability and a decrease in GABAergic levels in both M1 and thalamus. Lastly, we hypothesize that SICI at an IPI of 1 ms are different from IPIs at 2 and 4 ms, and that pulses at an IPI of 1 ms are related with GABA concentrations in M1.
METHODS
Ethical approval and informed consent
As part of two larger studies, this present study received ethical approval by the ethics committee of the faculty of Social and Behavioural Sciences of the University of Amsterdam. After participants were explained all procedures of each experiment, participants gave their written informed consent. Monetary compensation was given to participants upon completion of all experiments. All experiments performed met the standards set by the latest revision of the Declaration of Helsinki.
Participants
We recruited participants through the ‘Netherlands Sleep Registry’ (www.slaapregister.nl) and screened them via telephone interviews on their sleeping behaviour and general health. Exclusion criteria were other sleep disorders than primary insomnia, neurological or endocrinal disorders; claustrophobia, metal in the body, pregnancy and other factors excluding a person from MRI studies (see Appendix). Participants refrained from caffeine and alcohol consumption the evening before the experiments. Insomnia severity was scored on the Insomnia Severity Index (ISI; Morin & Barlow, 1993).
In this present thesis, the data collected by the author (‘WRAP’ study) was merged with the data collected from a different set on participants (‘VICI’ study), using the same setting and a very similar protocol. The two studies differ in that the TMS protocol in the WRAP study additionally included IPIs of 1 ms, and in that we additionally analyzed thalamic GABA levels in the WRAP study.
Of our 61 recruited participants, TMS data could not be collected for 5, because we were unable to locate the participant’s handknob, and TMS data could not be analyzed for 3 due to
technical issues (see Appendix). GABA levels of the motor cortex were not acquired for 3 participants, and GABA levels of the thalamus were not acquired for 4 participants. In total, we included 53 participants in our study, of who 13 were classified as suffering from moderate to severe insomnia (3 male, mean age = 55.7 ± 9.9 years, mean ISI = 18.5 ± 3.8) and 40 were healthy controls (9 male, mean age = 45.5 ± 16.3, mean ISI = 4.9 +- 4.10). Of these 53 participants, 38 were from the VICI study (6 males, mean age = 50.7 ± 13.9 years, mean ISI = 8.6 ± 7.7) and 15 from the WRAP study (6 males, mean age = 41.1 ± 17.6, mean ISI = 7.2 ± 5.5). These unequal sample sizes were accounted for by using a non-parametric statistical test that allows for different sample sizes.
TMS
Data acquisition
TMS data were acquired between 10:00 am and 3:00 pm, using a MagPro X100 stimulator (Medtronic, Minneapolis, Minnesota), connected to a figure-of-eight coil. We stimulated the hand knob in the left motor cortex in all participants. The motor responses in the right first dorsal interosseous (FDI), the hand muscle, were acquired via electromyography (EMG). For this, two electrodes were placed on the target muscle, two reference electrodes on the index finger and one ground electrode on the wrist. The data were sampled at 1024 Hz, amplified and recorded using a MOBI MINI amplifier and the Visor2 v2.0.1 software (Advanced Neuro Technology B.V.).
The location of the handknob was estimated as three centimeters lateral-anterior from the midpoint of the skull (Yousry et al., 1997). The TMS-coil was held in a 45-degree angle from the midsagittal line. By explorative search around that estimated location, the ‘hotspot’ was determined, i.e. the location in which the largest MEP amplitudes were found, and the TMS-coil was fixated. The resting-state motor threshold (MT) was defined as the minimum pulse
intensity necessary to elicit at least 5 MEPs of ~50 mV in the relaxed target muscle in 10 consecutive trials at estimated MT intensity.
Participants were placed in a massage chair and instructed to neither move nor speak for the rest of the experiment, which consisted of four distinct protocols. The first protocol consisted of 30 single pulses at MT strength, allowing to check that the stimulation intensity was at the actual MT. This was followed by a second protocol measuring SICI and ICF; consisting of 140 paired-pulses with varying inter-stimulus intervals (IPI: 1, 2, 4, 9, 12, 15, and 25 ms; 20 stimulations each), and 40 single-pulses at 120 % MT. With paired pulses, the first pulse had an intensity of 80 % MT, and the second pulse was at 120 % MT. The third protocol assessed LICI with 40 paired pulses at varying IPIs (50 ms, 100 ms, 150 ms, 200 ms, 250 ms). Here, both pulses were a 120 % of the MT. The trigger types of the second and third protocol appeared in pseudo-random order with delays varying between 3 and 7 seconds. The fourth protocol consisted of 30 single pulses, at 110, 130 and 150 % of the MT, acquiring a MEP recruitment curve.
First-level data analysis
The EMG-data were analyzed using a script written in Matlab R2014b (Mathworks, Natick, MA., USA) by the author (see Appendix). For the analysis of the MEP amplitude, peak-to-peak amplitudes were calculated for each trial. Following van der Werf et al. (2010), we calculated the amplitudes of SICI and ICF both as absolute values and relative to the amplitude of the 120 % MT single baseline pulse. LICI was calculated both in absolute terms and by expressing the amplitude of the MEP evoked by S2 as a percentage of the amplitude of S1. The slope of the MEP recruitment curve was calculated by applying a linear model on the median values. Due to outliers, we calculated the median of the MEP amplitude of every pulse type, instead of the mean.
MRS
Data acquisition
MRS data were acquired in the afternoon on the same day as TMS, using a 3T Philips Achieva MRI scanner and a 32-channel head coil along with a T1-weighted MR scan for registration (TR: 8.2 ms, TE: 3.8 ms, flip angle: 8 degrees).
Data analysis
MRS data were acquired with MEGA PRESS editing for GABA (Mullins et al., 2014), with TR: 2000ms, TE: 68 ms, flip angle: 180°, voxel size: 3 x 3 x 3 cm, and acquisition time: 11 minutes. The voxel was placed on the handknob of the motor cortex in the left hemisphere, and on the thalamus in the left hemisphere.
The data were pre-processed with Gannet, a Matlab-based toolkit for the quantitative analysis of GABA-edited MRS spectra (Edden et al., 2014). GABA values were given as a ratio to water in arbitrary units ([a. u.]). The T1-weighted structural image was segmented by using Partial Volume code for Philips MRS data by Nia Goulden and Paul Mullins (http://biu.bangor.ac.uk/projects.php.en). The GABA levels were corrected for grey matter volume within the voxel, by calculating the percentage of white and grey matter and cerebrospinal fluid in the voxel (Jensen et al., 2005).
TMS and MRS Second-level analysis
Statistical analyses were also conducted in Matlab R2014b. For the pairwise comparisons, the assumptions of parametric tests (homoscedasticity and normality) were checked. The assumptions were checked by Levene’s test and Kolmogorov-Smirnov test, of which the
results are not reported in this thesis, for the sake of readability. Since these assumptions were not met in all cases, the non-parametric Kruskal-Wallis test was applied, which is a one-way ANOVA by ranks. Likewise, the non-parametric test Kendall’s tau was used to test for correlations. Linear regression was done to predict GABA levels in the thalamus. For the first model, we included sex, age, MT and the interaction term of MT and age as predictors. In the second linear model only age and sex were included as predictors. The final model was derived by linear regression in a backward-eliminating step-wise fashion, including sex, age, MT and all interaction terms in the model.
RESULTS
For all pairwise comparisons, assumptions of parametric tests were checked; homoscedasticity was tested by conducting Levene’s test, and normality was checked with the Kolmogorov-Smirnov test. Since these assumptions were not met, the non-parametric Kruskal-Wallis test was applied. Likewise, the non-parametric test Kendall’s tau was used to test for correlations.
Insomnia scores
In contrast to van der Werf et al., we did not score insomnia according to the DSM criteria, but rather we classified insomniacs based on an ISI of 15 or higher, which is associated with moderate to severe insomnia. The ISI has been shown to be an adequate and consistent measure of insomnia (Bastien et al., 2000). Moreover, the ISI seems to classify insomnia in a stricter fashion than the DSM-V; we found in our data that the two classified differently in 11 participants, and in all of these cases, the DSM-V classified as insomniac, while ISI scores were in fact < 15. In order to clearly distinguish between good sleepers and true insomnia patients, we therefore decided to use the ISI. This might however entail discrepancies between the group differences in van der Werf’s study and this present study.
TMS
Main effect of TMS
In contrast to van der Werf’s findings, insomniacs and controls did not differ in MEP amplitudes following a 120 % MT baseline pulse (H(1) = 0.65, p = 0.420).
Visualizing the MEP amplitudes for SICI and ICF for all participants, a similar curve as van der Werf’s becomes apparent (fig. 1). To test if there was genuine inhibition or facilitation in the first three protocols, the Kruskal-Wallis test was applied for each double pulse to see if they differed from the MEP amplitudes either elicited by a baseline pulse at 120% MT (for SICI and ICF), or by S1 at 120% MT (for LICI) (table 1).
Paired pulses designed to induce ICF did not elicit a MEP larger than the baseline pulse. Similarly, pulses with long IPIs (50 to 200 ms) did not differ significantly from the conditioned pulse, with exception of pulses at IPIs of 100 ms. Consequently, only pulses at IPIs of 100 ms seemed to elicit genuine LICI.
Stimulating at increasing intensities indeed elicited stronger MEP responses. Comparing MEP amplitudes, we found significant larger MEPs between successive intensities (130% - 110%: H(1) = 36.19, p < 0.0001; 150% - 130%: H(1) = 8.8.5, p < 0.005). IPI (ms) 1 2 4 9 12 15 25 MEP size ( 7 V) 0 1000 2000 3000 4000 5000 6000 all participants
Fig. 1: MEP amplitude curve for SICI and ICF looks similar to van der Werf’s.
Absolute motor evoked potentials (MEPs) for short-interval intracortical inhibition (SICI; IPI = 1, 2, 4 ms) and intracortical facilitation (ICF; IPI = 9, 12, 15, 25 ms). Data points are of controls and insomniacs. Circles are median responses, error bars denote median absolute deviations.
Short-interval cortical inhibition
IPIs of 1 ms have been suggested to underlie different mechanisms than IPIs of 2 and 4 ms (Peurala et al., 2008; Ilić et al., 2002; Ziemann et al., 1996a). Comparing the median MEP amplitudes of pulses at IPIs of 1 ms with the median MEP amplitudes of pulses at 2 ms and 4 ms, we found significant differences (H(1) = 4.39, p < 0.05).
Insomniacs and controls of the WRAP and VICI study did not differ in the MEP amplitudes for SICI, neither in absolute values (median of IPIs of 2 ms and 4 ms; H(1) = 0.01, p = 0.918; fig. 2), nor in relation to the baseline pulse (H(1) = 0.23, p = 0.634; fig. 3). Similarly, expressing SICI as the median amplitudes for the IPIs of 1, 2 and 4 ms, insomniacs and control of the WRAP study did not differ significantly in their absolute MEP amplitudes
Table 1: Only SICI pulses and IPI = 100 ms differed from baseline / conditioned pulse.
Statistics for the Kruskal-Wallis test, comparing median MEP sizes with the baseline pulse (for SICI and ICF), or with the first, conditioned pulse (for LICI). Significant values are associated with a true inhibitory or facilitatory effect. MAD = median absolute
deviation. Sample sizes differ because IPI = 1 ms was only acquired for the WRAP study, and LICI could only be analyzed for the WRAP study.
IPI Type Median MAD H-value p-value n
1 SICI 45.76 18.60 16.36 < 0.000005* 15 2 SICI 140.63 73.35 24.76 < 0.000005* 53 4 SICI 117.70 77.71 36.22 < 0.000005* 53 9 ICF 684.93 486.96 0.45 0.501 53 12 ICF 626.33 510.71 0.18 0.674 53 15 ICF 619.81 465.13 0.25 0.620 53 25 ICF 550.96 407.75 0.04 0.842 53 50 LICI 33.85 25.71 1.00 0.319 15 100 LICI 24.44 22.89 5..42 < 0.05* 15 150 LICI 40.00 35.73 1.00 0.319 15 200 LICI 71.78 39.60 1.00 0.319 15 !
(H(1) = 0.0, p = 1.0). However, the difference between the two groups approached significance when expressed in relative terms to the baseline pulse (H(1) = 3.53, p = 0.060), but only if MEP amplitudes at IPIs of 1 ms were included; expressing relative SICI as relative MEP amplitudes at IPIs of 2 and 4 ms, there was no difference between the two groups (H(1) = 0.23, p = 0.634). Since pulses at an IPI of 1 ms have been suggested to underlie a different mechanism than other IPIs eliciting SICI (Stagg et al., 2011, Peurala et al., 2008), we separately tested for group differences in the 1 ms MEP amplitude; we found none in the absolute amplitude (H(1) = 0.75, p = 0.386), and one slightly approaching significance in the relative amplitude (H(1) = 2.53, p = 0.112).
Short-interval cortical facilitation
Insomniacs and controls did not differ in absolute MEP amplitudes at ICF (H(1) = 0.44, p = 0.508), and relative to baseline pulse (H(1) = 0.25, p = 0.620).!!
! IPI (ms) 1 2 4 9 12 15 25 MEP size ( 7 V) 0 1000 2000 3000 4000 5000 6000 Control Insomnia
Fig. 2: Insomniacs and controls did not differ in SICI (H(1) = 0.01, p = .918) and ICF (H(1) = 0.44, p = .508).
Absolute motor evoked potentials (MEPs) for short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF). Crosses are individual data points, circles are median MEP sizes, and error bars refer to median absolute deviation.
Long-interval cortical inhibition
The MEP amplitudes for the pulses designed to elicit LICI were equal for insomniacs and healthy controls (absolute values: H(1) = 0.71, p = .399; relative to conditioned pulse: H(1) = 0.52, p = .470; fig. 4). Visually inspecting fig. 4, relative MEP amplitudes at IPIs of 200 ms seemed to differ between insomniacs and controls; statistical analysis revealed a trend towards a significant difference (H(1) = 3.52, p = .060). Please note that for relative LICI MEP amplitudes, we only had access to the WRAP study.!!
IPI (ms) 1 2 4 9 12 15 25 MEP size (% 1.2 MT) 0 100 200 300 400 500 600 700 800 Control Insomnia
Fig. 3: Difference between insomniacs and controls in relation to baseline pulse approached significance for SICI (1, 2 and 4 ms) (H(1) = 3.53, p = .060).
Motor evoked potentials (MEPs) relative to baseline pulse for short-interval intracortical inhibition (SICI) and intracortical facilitation (ICF). Crosses are individual data points, circles are median MEP sizes, and error bars refer to median absolute deviation. The dashed line refers to the baseline pulse.
Stimulus Intensity (% MT) 110% 130% 150% MEP size ( 7 V) 0 1000 2000 3000 4000 500 6000 7000 8000 9000 Control Insomnia
Fig. 5: Insomniacs and controls did not differ in MEP recruitment size (H(1) = 0.91, p = .339).
Motor evoked potentials (MEPs) for increasing stimulus intensities. Crosses are individual data points, circles are median MEP sizes, and error bars refer to median absolute deviation.
Fig. 4: There was a trend towards a significant difference between insomniacs and controls at relative IPIs of 200ms (H(1) = 3.52, p = .060).
Motor evoked potentials (MEPs) in absolute values (left), and relative to baseline (right) for long-interval intracortical inhibition (LICI). Crosses are individual data points, circles are median MEP sizes, and error bars refer to median absolute deviation. Dashed line refers to the baseline pulse.
! IPI (ms) 50 100 150 200 MEP size ( 7 V) 0 500 1000 1500 2000 2500 3000 3500 4000 Control Insomnia IPI (ms) 50 100 150 200
MEP size (% conditioning pulse)
0 100 200 300 400 500 600 Control Insomnia
MEP recruitment curve
Comparing the insomniac group with the healthy controls, we found no significant difference in MEP recruitment slope (H(1) = 0.91, p = .339; fig. 5).
MRS
For the following analyses, please note that thalamic GABA levels were only acquired for 11 WRAP participants, while the GABA concentrations in the handknob were computed on 50 participants from both studies. Equally, correlations with the MEP amplitudes at IPI = 1 ms could only be conducted on the participants of the WRAP study (for handknob: n = 12; for thalamus: n = 11).
Main effect GABA
The GABA levels between the thalamus and the handknob differed significantly (H(1) = 22.35, p < .0001). The ratio between GABA concentrations in the thalamus vs. handknob does not differ between insomniacs and controls (H(1) = 4.22, p = .377).
Handknob Thalamus [GABA] (a.u.) 0 0.5 1 1.5 2 2.5 3 3.5
Fig. 6: GABA levels differed between handknob (motor cortex) and thalamus (H(1) = 22.35, p < .0001). GABA
GABA levels did not differ between insomniacs and controls in neither the thalamus (H(1) = 0.67, p = .414) nor the handknob (H(1) = 0.25, p = .619). As for trait-specific correlations, the amount of GABA in the thalamus differed between females and males (H(1) = 5.14, p < .05, Fig. 7); thus, females showed more GABA concentration in the thalamus than males. Thalamic GABA did not correlate with age (Kendall’s tau (11) = 0.224, p = .389). GABA concentrations in the handknob negatively correlated with age (Kendall’s tau (50) = -0.262, p < .01), but not with sex (Kendall’s tau (50) = 0.156, p = .190).
female male
[GABA thalamus] (a.u.)
0 0.5 1 1.5 2 2.5 3 3.5
Fig. 7: Thalamic GABA levels differed between females and males.
Females showed higher GABA concentrations in the thalamus than males (H(1) = 5.14, p < 0.05). Older people sh ow ed d ecre ased G ABA co nce ntr atio ns in th e ha ndknob of the mo to r co rtex (K en dal l’s tau (5 0) = -0.262, p < . 01). Dots ar e da ta points, the slo pe of th e d ash ed li ne is eq ual to th e co rrela tio n co effic ien t. Age 20 30 40 50 60 70
[GABA handknob] (a. u.)
1.4 1.6 1.8 2 2.2 2.4 2.6
GABA and TMS
Following Stagg et al. (2011), we tested if the MEP amplitudes of IPIs at 1 ms correlated with GABA levels. We computed non-significant effects for the GABA levels in the thalamus (Kendall’s tau (11) = -0.164, p = .542). Importantly, there was a trend effect for the negative correlation between GABA levels in the handknob and MEP amplitudes at 1 ms IPI ( ).
The MEP amplitudes at IPIs of 2 and 4 ms, suggested to elicit ‘true inhibition’, as opposed to IPIs of 1 ms (Peurala et al., 2008), did not correlate with GABA levels in the handknob (Kendall’s tau (50) = 0.093, p = .344) or the thalamus (Kendall’s tau (11) = -0.346, p = .165) either.
Equally, MEP amplitudes designed to elicit LICI did not correlate with GABA levels in the handknob (Kendall’s tau (50) = 0.130, p = .194) or thalamus (Kendall’s tau (11) = 0.164, p = .542).
No significant correlation was also computed between the slope of the MEP recruitment
curve and!GABA concentrations in neither handknob (Kendall’s tau (50) = 0.128, p = .225)
nor thalamus (Kendall’s tau (11) = -0.091, p = .761).
Lastly, we found that MT correlated with the amount of GABA in the thalamus (Kendall’s tau (11) = 0.587, p = .016), but it did not correlate with the GABA levels in the handknob (Kendall’s tau (50) = -0.131, p = .191). Correcting for the multiple comparisons we conducted on the thalamic GABA levels, we used Bonferroni corrections, and divided our significance level .05 by 5, which resulted in alpha = 0.01. With this, our corrected p-value is > .01.
To further investigate this relationship, we tested for trait-interactions with the motor threshold. We found that MT did not differ between the two sexes (H(1) = 1.58, p = .209), but age correlated significantly with MT intensity (Kendall’s tau (53) = 0.242, p < .05).
Insomniacs and controls did not differ in their MT (H(1) = 0.65, p = .419).
We used multiple regression to test if sex, age, MT or the interaction term of MT and age could predict the amount of GABA in the thalamus. We did not include the interaction between MT and sex, since they did not correlate with each other. The overall model fit was Adjusted R2
= 0.633 (p < .05). However, the variance of neither of the four terms by themselves explained the variance in thalamic GABA (see table 2, and appendix).
Motor Threshold
40 45 50 55 60 65
[GABA thalamus] (a. u.)
2 2.2 2.4 2.6 2.8 3 3.2 3.4
Fig. 9: Motor Threshold correlated with thalamic GABA concentrations.
Higher intensities required to elicit MEP responses positively correlated with thalamic GABA levels (Kendall’s tau (11) = 0.587, p = .016). Dots are data points, the slope of the dashed line is equal to the correlation coefficient.
Since MT and age were correlated, it is very likely that they both explained the same variance, and that this shared explained variance is split up between the two variables. To differentiate which of the two factors explains the most variance of thalamic GABA levels, we first included only age and sex in the model; this does not explain thalamic GABA levels (adjusted R2
= 0.3, p = .15). By doing stepwise (backward-eliminating) regression, including sex, age, MT and all interaction terms, the best fit explaining thalamic GABA concentrations was a model with only MT (ß = 0.200, p < .05) and sex (ß = 0.200, p = .054) as predictors (see fig. 10 and table 3). The overall model fit was Adjusted R2
= 0.655 (p < .01). Thus, we suggest that MT, together with sex, in fact is a predictor of thalamic GABA levels.
Beta estimate Beta Se t-value p-value
Intercept -4.22e-16 0.177 -2.38e-15 1
Motor threshold 0.575 0.200 2.882 0.020
Sex -0.451 0.200 -2-261 0.054
Table 2: Model containing MT, age, sex and MT:age predict thalamic GABA levels.
Whole model fit: adjusted R2 = 0.633 (p < 0.05). Beta values are z-standardized. Beta estimate Beta Se t-value p-value
Intercept 0.158 0.223 0.708 0.505
MT 0.188 0.376 0.500 0.635
Age -0.083 0.217 -0.381 0.716
Sex -0.428 0.218 -1.963 0.097
MT:Age -0.889 0.722 -1.231 0.264
Table 3: Model containing motor threshold and sex predict thalamic GABA levels.
Fig. 10: Sex and Motor Threshold predict thalamic GABA levels.
Left: Variance of thalamic GABA explained by Motor Threshold, controlling for Sex. Right: Variance of thalamic GABA explained by Sex, controlling for Motor Threshold. All values are z-standardized. Female is coded as 0, male is coded as 1. Whole model fit: adjusted R2 = 0.655 (p < .01).
Sex (z-score, adjusted for motor threshold)
-1.5 -1 -0.5 0 0.5 1 1.5
[GABA thalamus] (a. u.)
-1 -0.5 0 0.5
1 DataModel fit
Confidence bounds
Motor threshold (z-score, adjusted for sex)
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
[GABA thalamus] (a. u.)
-1.5 -1 -0.5 0 0.5 1
1.5 DataModel fit
DISCUSSION
Hyperarousal has been suggested to be a stable trait in insomnia, and multiple studies associate the insomnia-specific symptoms with hyperarousal, instead of merely the bad quality of sleep (van der Werf et al., 2010, Cano et al., 2008; Bonnet & Arand, 2010; Bonnet & Arand, 1997). Cortical excitability and ICF, as measured by TMS, have been shown to differ in insomnia patients, and thus possibly to reflect insomniac hyperarousal (van der Werf et al., 2010). GABA levels have been demonstrated to differ in insomnia (Winkelman et al., 2008; Waffors & Ebert, 2006) and hyperarousal (Centonze, 2005), and MRS-assessed GABA concentrations have been shown to be relatable to TMS measures of motor cortical excitability (Stagg et al., 2011). The aim of this study was to specifically relate TMS-assessed cortical excitability patterns with GABA levels in insomnia, and thereby to shed light on the underlying neurological mechanisms of hyperarousal in insomnia.
TMS and insomnia
We used a similar protocol as van der Werf et al. (2010). Visually inspecting the MEP amplitudes at SICI and ICF-eliciting IPIs (1 to 25 ms) (fig. 1), the curve appeared similar to van der Werf’s; with the exception of a smaller slope at ICF. By comparing the median amplitudes at each IPI with the baseline pulse, this impression was justified: there was a genuine SICI effect, but no ICF effect (see table 1); contrary to van der Werf’s findings. In accordance with van der Werf’s results, we did not find a group effect in absolute SICI amplitudes.
We also did not find a group effect in ICF, as van der Werf et al. did. This is possibly due to the fact that the pulses designed to elicit ICF in fact did not differ from the baseline pulse, i.e. they had no facilitatory effect.
Our experiments using the LICI protocol did not yield very promising results (fig. 4). Only pulses at IPI = 100 ms differed significantly from the baseline pulse, and only pulses at IPIs of 200 ms approached significant differences between insomniacs and healthy controls, but only in their relative values. It has been shown that inhibitory effects at IPIs > 50 ms are diminished with higher stimulus intensities in the first, unconditioned pulses; however, unconditioned pulses at 120% MT still elicited the maximum inhibitory effect (McNeil et al., 2011). McNeil et al. further reported that voluntary muscle contraction contaminated the LICI effect. Since the time lag between the first and the second pulse is too small to voluntarily elicit a muscle contraction, it could only be that participants were already tensed before. These are usually visible in the EMG response by smaller MEPs preceding or following the actual MEP response; visual inspection of our data did not give any evidence for increased voluntary muscle contraction. This all together is evidence for the theoretical effectiveness of our LICI protocol. Since there is still little known about the underlying mechanisms of LICI and its possible confounding factors, studies into this phenomenon are called for, in order to make claims about the null-effects in this present study.
Van der Werf et al. reported increased cortical excitability as measured by the 120 % MT baseline pulse in insomniacs. In our study on the other hand, insomniacs did neither show a higher MEP response to the baseline pulse, nor was their MEP recruitment curve characterized by a steeper slope. To our knowledge, no study so far has assessed MEP recruitment characteristics in insomnia.
Expressing the MEP amplitudes in SICI relative to the baseline pulse, the difference between insomniacs and controls approached significance, but only if MEP amplitudes at IPIs of 1 ms were included (fig. 3). We cautiously suggest that the inhibitory influence of the first, sub-threshold pulse on the second, supra-threshold pulse is different between the two groups,
and that this difference can be ascribed mainly to IPIs of 1 ms. Even though van der Werf et al. did not find this effect, we suggest that this is a very meaningful finding, since we showed that while all pulses at IPI = 1, 2 and 4 ms caused a genuine inhibitory effect, MEP amplitudes following pulses at IPIs of 1 ms differed from than those at 2 and 4 ms. This is evidence for our hypothesis that pulses at 1 ms IPI cause SICI by activating different cortical mechanisms than IPIs > 1 ms, which is in line with previous studies (Stagg et al., 2011, Peurala et al., 2008). More precisely, it has been proposed that the first pulse activates axons involved in the response to the second pulse; at intervals of 1 ms then, the neural response is still inhibited via the refractory period, and thus the response to the second pulse is diminished (Fisher et al., 2002). Moreover, when testing pulses from 1 ms to 4.4 ms in 0.2 ms increments, the most inhibitory effects, relative to the 120 % MT baseline pulse, were elicited by IPIs of 1 ms and 2.4 ms (Fisher et al., 2002). This might explain why we did not find any difference in SICI when not including IPIs at 1 ms. Further, even though IPIs at 1 ms alone did not differ significantly between the two groups, it has to be taken into consideration that these analyses were conducted on merely 15 participants; together with the just presented evidence from the literature, we suggest that a higher sample size might present us with significant differences.
We have thus found further evidence for a different inhibitory mechanism at IPIs of 1 ms; and, yet weaker, evidence for a difference between insomniacs and healthy individuals at this inhibitory mechanism. Since this effect is specifically apparent relative to the conditioned pulse, we hypothesize that in fact there is a difference in refractory period between the two groups.
MRS and insomnia
We showed that GABA levels differed between thalamus and handknob in both groups (fig. 6). This can be explained by the different distribution of neurotransmitters throughout the whole brain. Since the ratio between the GABA concentrations in these two brain areas did not differ between insomniacs and controls, we conclude that this difference is not associated with hyperarousal or insomnia.
GABA levels in either of our two seed areas also did not seem to differ between the two groups. Thus, we did not find support for our hypothesis that insomniacs exhibit less GABA in the thalamus. Evidence from previous studies does however not support these null-findings. The thalamus plays an important role in promoting sleep (Wafford & Sforza et al., 1995), GABA is highly abundant in the thalamus (Bateson, 2006), and one study showed a decrease in GABA levels, as measured in five brain areas, including the thalamus (Winkelman et al., 2008). But again, our null result has to be interpreted taking in account the low sample size of MRS-assessed GABA levels in the thalamus (n = 11).
TMS and MRS
We found some intriguing trait-specific correlations with GABA levels in both brain areas. Firstly, the amount of GABA in the handknob negatively correlated with age, and MT positively correlated with age. A positive correlation between age and resting MT has indeed been reported (Pitcher et al., 2003; Matsunaga et al., 1998). Furthermore, GABA concentrations have indeed been shown to decline with age in mesial frontal and parietal areas (Gao et al., 2013); and in an Alzheimer’s mice model, the impairment of GABAergic hilar interneurons have been correlated with age (Andrews-Zwilling et al., 2010). It has also been shown that the motor cortex excitability, as measured by TMS, declines with age (Oliviero et al., 2006); that inhibitory synapses in the motor cortex decline with increasing age (Christofi
et al., 2001); and that SICI, which relies on the excitability of GABAergic cortical circuits, also decreases with age (Peinemann et al., 2001). Importantly, MT has been shown to be independent of GABA (Ziemann et al., 1996a). Instead, MT can be influenced by drugs affecting sodium and calcium channels, indicating that MT reflects membrane excitability of corticospinal neurons (Hallett, 2000). Furthermore, at intensities close to the resting MT, the temporal summation of I-waves at motorneuron cell bodies has been reported to be an important determinant of MEP amplitude; it is possible that with increased age, less motorneurons are activated (Pitcher et al., 2003). Alternatively, the simultaneous activation of motorneurons could be impaired (Pitcher et al., 2003). This could be caused for instance by a frequency-dependent conduction block of the corticospinal fibres caused by an asynchronous firing of motorneurons (Mills, 1991).
We did find evidence for a positive link between thalamic GABA levels and MT. Sex also correlated with the amount of GABA in the thalamus, and since there was no relationship between sex and the MT, sex can be ruled out to be a covariate in the association between thalamic GABA and the MT. The thalamus is known to function as a major relay station for projections to sensorimotor areas (e.g. Darian-Smith et al., 2004). We speculate that increased GABA levels in the thalamus could influence the excitability of the handknob in the motor cortex.
General conclusions
This present study uncovered neurological mechanisms underlying hyperarousal in insomnia. It is suggested that the inhibitory influence of the unconditioned pulse on the conditioned pulse in SICI might differ between insomniacs and controls, and that this difference is mainly driven by pulses at IPI of 1 ms, which elicit a different – and stronger – inhibitory effect than pulses at 2 or 4 ms. More specifically, this differing inhibitory effect is suggested to be due to
the refractory period, which is therefore suggested to differ in insomniacs. There was a trend correlation effect between 1 ms pulses and GABA levels in the handknob. Considering the low sample size (n = 11), and the fact that Stagg et al. (2011) showed a significant correlation between exactly these pulses at 1 ms IPI and GABA levels in the handknob, we strongly suggest that our results will be significant, once we increase the sample size. However, no differences in GABA levels in the handknob have been found, which is why we cannot make the link between differences in refractory period and GABA levels for insomnia patients.
Apart from this, there was no effect of insomnia in the other TMS-assessed measure of cortical excitability. No significant differences between insomniacs and controls were found in GABA levels in the thalamus.
Increased GABA levels in the thalamus correlated with increased MTs, and MT together with sex significantly predicted thalamic GABA levels. However, the amount of GABA in the handknob did not directly correlate with MT. Due to correlations between GABA in the handknob and age, and correlations between age and MT, we cautiously suspect that higher concentrations of GABA in the thalamus are associated with higher resting MTs. We did not find evidence for the dependency of this decreased excitability of the motor cortex on GABA levels in the handknob.
We replicated van der Werf’s results only partly. There was a genuine SICI effect, and also an effect in the MEP recruitment slope. However, ICF and LICI pulses (besides IPIs of 100 ms) did not differ from the baseline pulse. We could not find any effect of ICF between insomniacs and healthy controls, very probably due to the non-facilitatory / non-inhibitory effect of these pulses. In part, these differences might be because of disparities in diagnosing insomnia: van der Werf classified insomnia according to DSM–IV criteria, while in this present study, insomnia was determined as scores > 15 on the ISI scale. However, since we
used a stricter tool to diagnose insomnia, the group effects were expected to be bigger than in van der Werf’s study.
REFERENCES
Andrews-Zwilling, Y., Bien-Ly, N., Xu, Q., Li, G., Bernardo, A., Yoon, S. Y., … Huang, Y. (2010). Apolipoprotein E4 causes age- and Tau-dependent impairment of GABAergic interneurons, leading to learning and memory deficits in mice. The Journal of Neuroscience!: The Official Journal of the Society
for Neuroscience, 30(41), 13707–13717. doi:10.1523/JNEUROSCI.4040-10.2010
Bastien, C. H., Vallières, A., & Morin, C. M. (2001). Validation of the insomnia severity index as an outcome measure for insomnia research. Sleep Medicine, 2(4), 297–307. doi:10.1016/S1389-9457(00)00065-4 Bateson, A. N. (2006). Further potential of the GABA receptor in the treatment of insomnia. Sleep Medicine, 7,
S3–S9. doi:10.1016/j.sleep.2006.03.001
Bonnet, M. H., & Arand, D. L. (1996). The consequences of a week of insomnia. II: Patients with insomnia.
Sleep, 21(4), 359–368.
Bonnet, M. H., & Arand, D. L. (1997). Hyperarousal and insomnia. Sleep Medicine Reviews, 1(2), 97–108. Bonnet, M. H., & Arand, D. L. (2010). Hyperarousal and insomnia: state of the science. Sleep Medicine Reviews,
14(1), 9–15. doi:10.1016/j.smrv.2009.05.002
Bpharm, S. C., Wilson, S., Dm, D. N., Chaplin, S., Wilson, S., & Nutt, D. (2013). n DRUG POINTS Z drugs!: their properties and use in treating insomnia, (May), 32–33.
Cano, G., Mochizuki, T., & Saper, C. B. (2008). Neural circuitry of stress-induced insomnia in rats. The Journal
of Neuroscience!: The Official Journal of the Society for Neuroscience, 28(40), 10167–10184.
doi:10.1523/JNEUROSCI.1809-08.2008
Centonze, D., Palmieri, M. G., Boffa, L., Pierantozzi, M., Stanzione, P., Brusa, L., … Tor, U. (2005). Cortical hyperexcitability in post-traumatic stress disorder secondary to minor accidental head trauma!: a neurophysiologic study, 30(2), 127–132.
Christofi, F. L., Zhang, H., Yu, J. G., Guzman, J., Xue, J., Kim, M., … Cooke, H. J. (2001). Age-related decline of presumptive inhibitory synapses in the sensorimotor cortex as revealed by the physical disector. Journal
of Comparative Neurology, 439(1), 65–72. doi:10.1002/cne.1335
Di Lazzaro, V., Oliviero, a, Profice, P., Pennisi, M. a, Pilato, F., Zito, G., … Tonali, P. a. (2003). Ketamine increases human motor cortex excitability to transcranial magnetic stimulation. The Journal of Physiology,
547, 485–496. doi:10.1113/jphysiol.2002.030486
Di Lazzaro, V., Rothwell, J. C., Oliviero, a, Profice, P., Insola, a, Mazzone, P., & Tonali, P. (1999). Intracortical origin of the short latency facilitation produced by pairs of threshold magnetic stimuli applied to human motor cortex. Experimental Brain Research. Experimentelle Hirnforschung. Experimentation Cerebrale,
129, 494–499. doi:10.1007/s002210050919
Edden, R. a E., Puts, N. a J., Harris, A. D., Barker, P. B., & Evans, C. J. (2013). Gannet: A batch-processing tool for the quantitative analysis of gamma-aminobutyric acid-edited MR spectroscopy spectra. Journal of
Magnetic Resonance Imaging, 1452, 1445–1452. doi:10.1002/jmri.24478
Fisher, R. J., Nakamura, Y., Bestmann, S., Rothwell, J. C., & Bostock, H. (2002). Two phases of intracortical inhibition revealed by transcranial magnetic threshold tracking. Experimental Brain Research, 143(2), 240–248. doi:10.1007/s00221-001-0988-2
Gao, F., Edden, R. a E., Li, M., Puts, N. a J., Wang, G., Liu, C., … Barker, P. B. (2013). Edited magnetic resonance spectroscopy detects an age-related decline in brain GABA levels. NeuroImage, 78, 75–82. doi:10.1016/j.neuroimage.2013.04.012
Hallett, M. (2000). Transcranial magnetic stimulation and the human brain. Nature, 406(6792), 147–150. doi:10.1038/35018000
Ilić, T. V, Meintzschel, F., Cleff, U., Ruge, D., Kessler, K. R., & Ziemann, U. (2002). Short-interval paired-pulse inhibition and facilitation of human motor cortex: the dimension of stimulus intensity. The Journal
of Physiology, 545, 153–167. doi:10.1113/jphysiol.2002.030122
Jensen, J. E., de B. Frederick, B., & Renshaw, P. F. (2005). Grey and white matter GABA level differences in the human brain using two-dimensional, J-resolved spectroscopic imaging. NMR in Biomedicine, 18(8), 570–576. doi:10.1002/nbm.994
Lang, E. J. (2002). GABAergic and glutamatergic modulation of spontaneous and motor-cortex-evoked complex spike activity. Journal of Neurophysiology, 87(4), 1993–2008. doi:10.1152/jn.00477.2001
Lanza, G., Cantone, M., Lanuzza, B., Pennisi, M., Bella, R., Pennisi, G., & Ferri, R. (2014). Distinctive patterns of cortical excitability to transcranial magnetic stimulation in obstructive sleep apnea syndrome, restless legs syndrome, insomnia, and sleep deprivation. Sleep Medicine Reviews, 19, 39–50.
doi:10.1016/j.smrv.2014.04.001
Liepert, J., Schwenkreis, P., & Tegenthoff, M. (1997). The glutamate antagonist Riluzole suppresses intracortical facilitation. Journal of Neural Transmission, 104, 1207–1214.
Matsunaga, K., Uozumi, T., Tsuji, S., & Murai, Y. (1998). Age-dependent changes in physiological threshold asymmetries for the motor evoked potential and silent period following transcranial magnetic stimulation.
Electroencephalography and Clinical Neurophysiology, 109(6), 502–507.
doi:10.1016/S1388-2457(98)00020-0
Mattay, V. S., Fera, F., Tessitore, a, Hariri, a R., Das, S., Callicott, J. H., & Weinberger, D. R. (2002). Neurophysiological correlates of age-related changes in human motor function. Neurology, 58(4), 630– 635. doi:Not available
McDonnell, M. N., Orekhov, Y., & Ziemann, U. (2006). The role of GABAB receptors in intracortical inhibition in the human motor cortex. Experimental Brain Research, 173, 86–93. doi:10.1007/s00221-006-0365-2 McNeil, C. J., Martin, P. G., Gandevia, S. C., & Taylor, J. L. (2011). Long-interval intracortical inhibition in a
human hand muscle. Experimental Brain Research, 209(2), 287–297. doi:10.1007/s00221-011-2552-z Mills, K. R. (1991). Magnetic brain stimulation: A tool to explore the action of the motor cortex on single human
spinal motoneurones. Trends in Neurosciences, 14(9), 401–405. doi:10.1016/0166-2236(91)90029-T Morgan, P. T., Pace-Schott, E. F., Mason, G. F., Forselius, E., Fasula, M., Valentine, G. W., & Sanacora, G.
(2012). Cortical GABA levels in primary insomnia. Sleep, 35(6), 807–14. doi:10.5665/sleep.1880 Morin, C. M., & Barlow, D. H. (1993). Insomnia: Psychological assessment and management (Vol. 104).
Guilford Press New York.
Morin, C. M., LeBlanc, M., Daley, M., Gregoire, J. P., & Mérette, C. (2006). Epidemiology of insomnia: Prevalence, self-help treatments, consultations, and determinants of help-seeking behaviors. Sleep
Medicine, 7(2), 123–130. doi:10.1016/j.sleep.2005.08.008
Mullins, P. G., McGonigle, D. J., O’Gorman, R. L., Puts, N. a J., Vidyasagar, R., Evans, C. J., & Edden, R. a E. (2014). Current practice in the use of MEGA-PRESS spectroscopy for the detection of GABA.
NeuroImage, 86, 43–52. doi:10.1016/j.neuroimage.2012.12.004
Nakamura, H., Kitagawa, H., Kawaguchi, Y., & Tsuji, H. (1997). Intracortical facilitation and inhibition after transcranial magnetic stimulation in conscious humans, 817–823.
Ohayon, M. M. (2002). Epidemiology of insomnia: What we know and what we still need to learn. Sleep
Medicine Reviews, 6(2), 97–111. doi:10.1053/smrv.2002.0186
Oliviero, a., Profice, P., Tonali, P. a., Pilato, F., Saturno, E., Dileone, M., … Di Lazzaro, V. (2006). Effects of aging on motor cortex excitability. Neuroscience Research, 55(1), 74–77.
doi:10.1016/j.neures.2006.02.002
Peinemann, a, Lehner, C., Conrad, B., & Siebner, H. R. (2001). Age-related decrease in paired-pulse intracortical inhibition in the human primary motor cortex. Neuroscience Letters, 313(1-2), 33–36. Peurala, S. H., M. Müller-Dahlhaus, J. F., Arai, N., & Ziemann, U. (2008). Interference of short-interval
intracortical inhibition (SICI) and short-interval intracortical facilitation (SICF). Clinical Neurophysiology,
119, 2291–2297. doi:10.1016/j.clinph.2008.05.031
Pitcher, J. B., Ogston, K. M., & Miles, T. S. (2003a). Age and sex differences in human motor cortex input-output characteristics. The Journal of Physiology, 546(Pt 2), 605–613. doi:10.1113/jphysiol.2002.029454 Riemann, D., Spiegelhalder, K., Feige, B., Voderholzer, U., Berger, M., Perlis, M., & Nissen, C. (2010). The
hyperarousal model of insomnia: A review of the concept and its evidence. Sleep Medicine Reviews, 14(1), 19–31. doi:10.1016/j.smrv.2009.04.002
Rothwell, J. C., Day, B. L., Thompson, P. D., & Kujirai, T. (2009). Short latency intracortical inhibition: one of the most popular tools in human motor neurophysiology. The Journal of Physiology, 587(Pt 1), 11–2. doi:10.1113/jphysiol.2008.162461
Salas, R. E., Galea, J. M., Gamaldo, A. a, Gamaldo, C. E., Allen, R. P., Smith, M. T., … Celnik, P. a. (2014). Increased use-dependent plasticity in chronic insomnia. Sleep, 37, 535–44. doi:10.5665/sleep.3492 Saper, C. B., Scammell, T. E., & Lu, J. (2005). Hypothalamic regulation of sleep and circadian rhythms. Nature,
437(7063), 1257–63. doi:10.1038/nature04284
Schwenkreis, P., Witscher, K., Janssen, F., Addo, A., Dertwinkel, R., Zenz, M., … Tegenthoff, M. (1999). Influence of the N-methyl-D-aspartate antagonist memantine on human motor cortex excitability.
Neuroscience Letters, 270, 137–140. doi:10.1016/S0304-3940(99)00492-9
Sforza, E., Montagna, P., Tinuper, P., Cortelli, P., Avoni, P., Ferrillo, F., … Lugaresi, E. (1995). Sleep-wake cycle abnormalities in fatal familial insomnia. Evidence of the role of the thalamus in sleep regulation.
Electroencephalography and Clinical Neurophysiology, 94(6), 398–405.
doi:10.1016/0013-4694(94)00318-F
Sivertsen, B., Omvik, S., Pallesen, S., Bjorvatn, B., Havik, O., Kvale, G., … Nordhus, I. H. (2006). Cognitive Behavioral Therapy vs Zopiclone for Treatment of Chronic Primary Insomnia. Journal of the American
Medical Association, 295(24), 2851–2858.
Soares, D. P., & Law, M. (2009). Magnetic resonance spectroscopy of the brain: review of metabolites and clinical applications. Clinical Radiology, 64(1), 12–21. doi:10.1016/j.crad.2008.07.002
Stagg, C., Bestmann, S., Constantinescu, A., Moreno, L., Allman, C., Mekle, R., … Rothwell, J. (2011). Relationship between physiological measures of excitability and levels of glutamate and GABA in the human motor cortex., 23, 5845–5855. doi:10.1113/jphysiol.2011.216978
Stagg, C. J. (2014). Magnetic Resonance Spectroscopy as a tool to study the role of GABA in motor-cortical plasticity. NeuroImage, 86, 19–27. doi:10.1016/j.neuroimage.2013.01.009
Stagg, C. J., Bachtiar, V., & Johansen-Berg, H. (2011). What are we measuring with GABA magnetic resonance spectroscopy? Communicative & Integrative Biology, 4(February 2015), 573–575.
Stagg, C. J., Best, J. G., Stephenson, M. C., O’Shea, J., Wylezinska, M., Kincses, Z. T., … Johansen-Berg, H. (2009). Polarity-sensitive modulation of cortical neurotransmitters by transcranial stimulation. The Journal
of Neuroscience!: The Official Journal of the Society for Neuroscience, 29(16), 5202–5206.
doi:10.1523/JNEUROSCI.4432-08.2009
Strafella, a P., & Paus, T. (2001). Cerebral blood-flow changes induced by paired-pulse transcranial magnetic stimulation of the primary motor cortex. J Neurophysiol, 85, 2624–2629. Retrieved from
http://www.jn.physiology.org/cgi/content/full/85/6/2624\nhttp://www.jn.physiology.org/cgi/content/abstra ct/85/6/2624\nhttp://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation &list_uids=11387406
Tokimura, H., Ridding, M. C., Tokimura, Y., Amassian, V. E., & Rothwell, J. C. (1996). Short latency facilitation between pairs of threshold magnetic stimuli applied to human motor cortex.
Electroencephalography and Clinical Neurophysiology - Electromyography and Motor Control, 101, 263–
272. doi:10.1016/0924-980X(96)95664-7
Van der Werf, Y. D., Altena, E., van Dijk, K. D., Strijers, R. L. M., De Rijke, W., Stam, C. J., & van Someren, E. J. W. (2010). Is disturbed intracortical excitability a stable trait of chronic insomnia? A study using transcranial magnetic stimulation before and after multimodal sleep therapy. Biological Psychiatry,
68(10), 950–5. doi:10.1016/j.biopsych.2010.06.028
Wafford, K. a, & Ebert, B. (2006). Gaboxadol--a new awakening in sleep. Current Opinion in Pharmacology,
6(1), 30–6. doi:10.1016/j.coph.2005.10.004
Winkelman, J. W., Buxton, O. M., Jensen, J. E., Benson, K. L., Connor, S. P. O., Wang, W., & Renshaw, P. F. (2008). Reduced Brain GABA in Primary Insomnia!: Preliminary Data from 4T Proton Magnetic Resonance Spectroscopy ( 1H-MRS ). Sleep, 31(11), 1499:1506.
Yousry, T. a., Schmid, U. D., Alkadhi, H., Schmidt, D., Peraud, a., Buettner, a., & Winkler, P. (1997). Localization of the motor hand area to a knob on the precentral gyrus. A new landmark. Brain, 120(1), 141–157. doi:10.1093/brain/120.1.141
Ziemann, U., Chen, R., Cohen, L. G., & Hallett, M. (1998). Dextromethorphan decreases the excitability of the human motor cortex. Neurology, 51, 1320–1324. doi:10.1212/WNL.51.5.1320
Ziemann, U., Lönnecker, S., Steinhoff, B. J., & Paulus, W. (1996a). Effects of antiepileptic drugs on motor cortex excitability in humans: a transcranial magnetic stimulation study. Annals of Neurology, 40, 367– 378. doi:10.1002/ana.410400306
Ziemann, U., Lönnecker, S., Steinhoff, B. J., & Paulus, W. (1996b). The effect of lorazepam on the motor cortical excitability in man. Experimental Brain Research. Experimentelle Hirnforschung.
APPENDIX
Exclusion criteriaNeurological disorders excluding potential participants from our stusy were migraine, epilepsy, congenital epilepsy in first or second degree family members. Endocrinal disorders used as exclusion criteria were e.g. thyroid or adrenal dysfunction. Psychological disorders such as depression, current use of anti-depressants, placement of a stent during an angioplasty surgery also excluded from participation. As for MRI exclusion criteria, people having metal material inside the body, being (ex-) metalworker, being claustrophobic, and being pregnant were excluded. As for the sleep criteria, shift work, parasomnia (i.e. sleep apnea, restless leg syndrome, and/or narcolepsy) were exclusion criteria.
During the acquisition and analyses of TMS data, we excluded 4 participants because we were not able to find the handknob, 2 participants because the amplifier ceased working, and MEP responses were scattered in time, and 2 additional participants showed MEP responses, which were scattered in time due to other reasons than amplifier failure.
TMS-Data Analysis
The TMS data was analyzed with the help of Matlab 2014 b. First, the EMG data was divided up into smaller time windows, matching the specific TMS pulse preceding the MEP response. We wrote an algorithm which detected the peaks in the EMG data, corresponding to the MEP itself. By taking the sum between the voltage at trough and at peak, we calculated the
amplitude of the MEP. The correct performance of this peak detection algorithm was checked by visual inspection for every trial of every participant.
In first-level analysis, we visually inspected the data of every protocol for each participant for noise. Based on this, the data of two participants was excluded, one because the amplifier ran out of battery during testing and the timing of the EMG data was incorrect; and the other
one because the first protocol showed hardly no responses in any of the 30 trials, suggesting that we had missed the handknob, or that the threshold was falsely determined.
In second-level analysis, we divided the participants up into control group and insomniac group, computed medians for each IPI and each participant, and carried out statistical tests, using Matlab R 2014b.
Model explaining thalamic GABA level
!
Fig. 10: MT, Age, Sex and the interaction of MT and age together predict thalamic GABA levels.
All values are z-standardized. Female is coded as 0, male is coded as 1. MT is motor threshold; MT:age is the interaction effect of MT and age.
The whole model fit was significant (p < 0.05, adjusted R2 = 0.633). The variance of neither of the four terms by themselves explained the variance in thalamic GABA (see table 2).
Sex (z-score, adjusted for age, MT and MT:age)-1.5 -1 -0.5 0 0.5 1 1.5
[GABA thalamus] (a. u.)
-1.5 -1 -0.5 0 0.5 1 1.5 Data Model fit Confidence bounds
Age (z-score, adjusted for sex, MT and MT:age)
-1 -0.5 0 0.5 1 1.5
[GABA thalamus] (a. u.)
-1 -0.5 0 0.5 1 Data Model fit Confidence bounds
MT:age (z-score, adjusted for sex, age and MT)-1 -0.5 0 0.5 1 1.5
[GABA thalamus] (a. u.)
-1 -0.5 0 0.5
1 DataModel fit
Confidence bounds MT (z-score, adjusted for sex, age and MT:age)-1 -0.5 0 0.5 1 1.5
[GABA thalamus] (a. u.)
-1 -0.5 0 0.5 1 Data Model fit Confidence bounds
