9.2.1.1 Studies Targeting the Motor Cortex

Even single transcranial magnetic stimuli, when applied to the human motor cortex, can induce a high-frequency repetitive discharge of corticospinal neurons [8]. These waves of activation may originate from the direct activation of the axons of the pyramidal tract neurons (PTNs) at the initial segment, i.e., the so-called “direct” waves or “D” waves, or from indirect, transsynaptic activation of the PTNs mediated by intracortical circuits, i.e., the so-called “indirect” waves or “I” waves [9]. The precise mechanisms at the base of the corticospinal activity induced by TMS remain speculative due to the complexity of the interactions between the experimentally induced magnetic field and the cortical circuits composed of multiple layers of excitatory and inhibitory interneurons and axons of various size, location, orientation, and function [1]. There is evidence that different factors such the orientation of the coil and the intensity of stimulation remarkedly influence the motor cortical response to a magnetic stimulus, likely as a result of the activation of different neuronal populations and/or axons [10]. Three main electrophysiological parameters can be assessed by means of single-pulse TMS, i.e., the motor threshold (MT), the input–output (I–O) curve, and the cortical silent period (CSP). MT is defined as the minimum intensity of TMS capable to evoke a peripheral motor response in at least 50% of trials, and it can be obtained either when the target muscle is at rest (resting MT, or RMT) or during active moderate contraction (active MT, or AMT) [11]. This parameter is greatly variable across individuals but also rather constant in a given subject [12]. Several factors affect the MT, including non-modifiable anatomical characteristics such as the scalp–cortex distance [13], or factors influencing the ion channels conductivity in the neuronal membranes. The latter notion is mainly based on finding that sodium and calcium channel blockers can increase MT values while antiepileptic drugs with neurotransmitter properties have not such an effect [14]. The search for possible differences in the MT between migraineurs and control subjects has yield very inconsistent results. Although some studies have reported lower [15, 16] or higher [17–20] MT values in migraine, most of the authors have found no significant differences [21–24]. Clinical differences between the patients’ groups could explain these discrepancies. For example, in a recent work by Cortese et al. [25] in patients with migraine without aura evaluated between attacks, cyclical changes in the RMT values were observed in relation to the time elapsed from the last migraine attack. In particular, a negative correlation between the RMT and the number of days since the last attack was recorded.

The input–output (I–O) curve, or recruitment curve, has been proven to be a more sensitive electrophysiological measure to detect changes in intracortical excitability with respect to the MT. In this TMS protocol, stimulation over the motor cortical “hot spot” for eliciting responses in a target muscle is given with increasing stimulus intensities, resulting in larger motor-evoked potentials (MEPs) until a plateau is observed in the amplitude of the MEPs. The growth of the MEPs with higher intensities of stimulation is thought to reflect a growing recruitment of excitatory intracortical circuits that generate a series of indirect waves of transsynaptic activation of the pyramidal tract neurons [26–28]. Different studies have shown a steeper slope of the I–O curve in patients suffering from migraine with and without aura as compared to healthy subjects [15, 16, 29]. This finding of increased motor cortical excitability could be interpreted as due to a more pronounced reverberating activation of facilitatory intracortical circuits, leading to an excitation/inhibition imbalance within the motor cortex. In support to this hypothesis is the finding that, in patients with MwA, treatment with levetiracetam can decrease the slope of the I–O curve [29]. Indeed, it is known that levetiracetam inhibits glutamatergic neurotransmission [30, 31] without affecting neither activity of inhibitory intracortical neurons nor measures of spinal and peripheral excitability [32, 33]. An interesting finding by Van der Kamp et al. [16] was that MEP amplitudes positively correlated with the attack frequency. This datum could explain findings by Maertens de Noordhout [19], who failed to find increased MEP amplitudes in MwA patients with a low attack frequency.

Finally, CSP represents another attractive measure easily evaluable by means of single magnetic stimuli. CSP refers to a transient interruption of the voluntary muscle contraction induced by transcranial stimulation of the contralateral motor cortex [34]. The duration of the CSP is a measure of intracortical inhibition mediated by activation of metabotropic gamma-aminobutyric-acid-B receptors (GABA_BR) which thus reflects activity of GABAB interneurons that synapse on pyramidal neurons [35–37]. Most of the studies have shown a reduced CSP duration in migraineurs in support of the hypothesis that an inhibitory circuits dysfunction could lead to cortical hyperresponsivity [15, 38–42]. However, this datum was not confirmed by other authors [23, 24], and it remains matter of debate whether CSP reduction represents a primary dysfunction of cortical inhibitory interneurons or is expression of a reduced preactivation by hypoactive thalamocortical loops [39, 41]. Interestingly, a negative correlation between CSP duration and frequency of the migraine attacks was found by Aurora et al. [38]. Moreover, in a study by Currà et al. [40] in patients with chronic headache related to medication overuse (MOH), CSP duration was found to positively correlate with monthly intake of nonsteroid anti-inflammatory drugs (NSAIDs) and combination drugs, while a negative correlation with monthly tablets intake was observed in triptan overusers. These findings lead to a more direct consideration of the relationship between activity of GABAB interneurons, degree of clinical impairment, and the effects of medications on the brain.

9.2.1.2 Studies Targeting the Visual Cortex

When a magnetic stimulus is applied to the occipital visual cortex, a subject can experience the perception of artificial lights or flashes, known as “phosphenes.” In individual subjects, TMS-induced phosphenes can be used to investigate visual cortical excitability by determining the threshold intensity of stimulation at which phosphenes are elicited on about half of trials (i.e., the “phosphene threshold,” PT) [43]. Aurora et al. [44] were the first to report a greater propensity to perceive phosphenes and lower PT values in migraine with aura patients with respect to controls, so providing the first neurophysiological correlate of visual cortex hyperexcitability. Later, other authors found similar results also in migraine without aura patients [23, 45, 46], but again some authors failed to confirm them [17, 47] or even reported increased PT values in the patients’ group [21]. Among the potential causes of this discrepancy, the most relevant probably refers to the type of coil employed. In a recent meta-analysis by Brigo et al. [48], it was found that in both patients with migraine with and without aura, significant lower PT values with respect to healthy subjects may be detected only by using a circular instead of a figure-of-eight coil. Compared to a figure-of-eight coil that is more suitable for focal stimulation, a circular coil stimulates larger cortical areas and could also generate a stronger electric current [49]. Both of these factors could contribute to make the circular coil more suitable to highlight differences in visual cortical excitability between patients and controls.

Other neurophysiological techniques have been employed in migraine patients to investigate excitability of the visual cortex. Mulleners et al. [50] first tested in migraine with aura patients a paradigm in which a timed magnetic stimulus applied over the visual cortex immediately after initiation of a visual stimulus can reduce accuracy of visual perception [51]. They found a reduced “magnetic suppression of perceptual accuracy” (MSPA) that was interpreted as consequence upon a defective activation of intracortical inhibitory circuits of the visual cortex. By using the same technique, Aurora et al. [52] confirmed these results in episodic migraineurs and also revealed an even higher cortical excitability level in patients with probable chronic migraine, so suggesting that this could contribute to the increased attack frequency. Finally, Battelli et al. [53] aimed to examine visual cortical excitability by applying TMS over an extrastriate area of the brain (area V5), which is involved in visual perception of motion. These authors showed a reduced threshold for the induction of “moving phosphenes” in migraine patients, thus indicating that hyperexcitability of the visual cortex goes beyond visual area V1, as it may also involve the association visual cortex.

9.2.2.1 Studies Targeting the Motor Cortex

The paired-pulse TMS paradigm represents another objective and validated method to investigate the excitability of the primary motor cortex [54]. In this approach, researchers evaluate the modulatory effect of a first conditioning pulse, usually given below the resting MT, on the amplitude of the MEP elicited by a second suprathreshold test stimulus. This effect strictly depends on the interstimulus interval (ISI) used. At very short intervals, less than 5 ms, there is MEP inhibition, known as “short intracortical inhibition” (SICI). Instead, by using ISIs between 8 and 30 ms, an MEP potentiation, referred as “intracortical facilitation” (ICF), can be observed. At longer ISIs of 50–400 ms, MEP inhibition can occur again (“long intracortical inhibition,” LICI), but only by applying a suprathreshold conditioning pulse, usually equal to 120% or 130% of the resting MT [55]. SICI reflects activation of inhibitory, GABAergic intracortical neurons within the primary motor cortex, and in particular, it is considered as a marker of GABAA activity [56]. More controversial is the interpretation of the effects observed at longer ISIs. As well as for SICI, ICF is widely believed to originate within the primary motor cortex, as the sum of a facilitatory component from NMDA receptors and an inhibitory component from GABAA receptors [14, 57]. However, subcortical mechanisms have additionally been considered to play a role [58], and accordingly, corticospinal effects and facilitation of H-reflexes have been also recorded after subthreshold TMS [59]. There is even less knowledge regarding LICI. The mechanism is complex since the effect of a suprathreshold conditioning stimulus goes beyond modifying activity within motor cortex, inducing corticospinal activity which can change the excitability of spinal circuits as well. Moreover, growing evidence suggests that inhibition observed at different relatively long ISIs involves, at least partially, different processes [60]. As far as we know, at ISIs less than 50 ms, LICI might mainly reflect the influence of spinal mechanisms, while inhibitory effects seen at longer ISIs (i.e., more than 100 ms) could be mainly cortical in nature and mediated by the activation of GABAB receptors [61, 62].

The use of the paired-pulse TMS paradigm is especially attractive, if we consider that a functional imbalance between inhibitory and excitatory intracortical circuits has been considered of major importance for migraine pathophysiology [63, 64]. As regards intracortical facilitation, most previous studies have found similar ICF values in interictal migraineurs compared to healthy controls [17, 24, 42, 65, 66]. However, according to the evidence that in migraine different methods of cortical stimulation can induce different responses in relation to the stimulation intensity used [63], some authors found that an increased ICF may be recorded in migraine only by using precise stimulation intensities for either the conditioning or the test stimulus. Indeed, Siniatchkin et al. [67] found increased ICF values in migraineurs compared to controls by using a suprathreshold intensity for the conditioning stimulus. Conversely, Cosentino et al. [68] showed that when a subthreshold intensity for the conditioning stimulus was used, an increased ICF could be recorded in migraineurs only by applying the test stimulus at a relatively low stimulation intensity (i.e., 110% of the RMT).

As for ICF, different stimulation parameters have been adopted by various authors for assessing SICI, thus leading to heterogeneity in results among studies. Indeed, on the one hand, reduced SICI values have been reported by Brighina et al. [65, 66], and more recently by Neverdahl et al. [42], thus supporting the hypothesis of an abnormal GABAergic inhibition within the migraine motor cortex. On the other hand, no statistically significant differences both as regards SICI [17, 24, 68] and LICI values [67, 68] were found by other authors.

9.2.2.2 Studies Targeting the Visual Cortex

To our knowledge, only Gerwig et al. [69] have used a paired-pulse TMS protocol to test visual cortical excitability in migraine without aura patients so far. These authors used a figure-of-eight coil and applied both single and paired magnetic pulses to the visual cortex of patients and controls, showing that significant reduced PT values could be found in migraineurs only when the paired-pulse protocol was used. Thus, this latter paradigm could be more sensitive in detecting changes in visual cortical excitability compared to more conventional methods used to assess the PT.

9.2.3.1 Studies Targeting the Motor Cortex

Repetitive transcranial magnetic stimulation (rTMS) has been proven capable to induce long-term neuroplasticity effects in humans under normal and pathological conditions, thus opening interesting perspectives in its potential use as treatment for various neurological and psychiatric disorders. The physiological bases underlying modulation induced by rTMS have not been completely elucidated. Many arguments support the concept that the rTMS aftereffects are mediated by mechanisms resembling long-term potentiation (LTP) and long-term depression (LTD) described in animal experiments [70, 71]. In vitro, LTP can be induced by trains of high-frequency stimulation, whereas LTD occurs after relatively long low-frequency stimulation periods. In a similar manner, when applied in healthy human subjects, high- and low-frequency rTMS sessions can induce, respectively, a transient facilitation or inhibition of cortical excitability [72, 73]. Changes in synaptic strength observed in animal in vitro experiments, as well as cortical excitability shifts induced in humans in vivo by rTMS, can be divided into two phases. A short phase, known as early LTP or LTD, refers to changes lasting for only 30–60 min. A long phase, called late LTP or LTD, can be observed when stimulations are repeated over longer periods and are due to modifications of protein synthesis [71].

Clear evidence has been provided that the migraine motor cortex responds to various rTMS paradigms differently from that observed in the healthy subjects. In the normal subjects, rTMS given at slow rates, between 0.2 and 1 Hz, induces a decrease in motor cortical excitability, while an opposite facilitatory effect is induced by rTMS at faster rates, i.e., ≥5 Hz [71, 74]. Paradoxical responses to both low- and high-frequency rTMS have been shown in patients with migraine. As regards low-frequency rTMS, a paradoxical facilitatory effect on intracortical facilitation has been observed after single 1 Hz rTMS sessions [65]. This was interpreted as possible consequence of dysfunction of the GABAergic intracortical circuits that could not be upregulated by low-frequency rTMS. However, another possible explanation is that in migraine, hyperresponsive glutamatergic circuits could overreact to 1 Hz rTMS, so reaching the threshold for inducing LTP-like mechanisms. This hypothesis agrees with the evidence that both LTD and LTP synaptic plasticity can be triggered by NMDA-dependent depolarization of the postsynaptic membrane rather than by long-lasting changes in the activity of the GABAergic intracortical circuits [75].

Abnormal responses to rTMS have been shown in migraine also when applying high-frequency rTMS. In the normal subjects, brief trains of 5 Hz rTMS applied to the M1 induce a progressive and short-lasting MEP potentiation only when using suprathreshold stimulation intensities equal or up to 120% of the RMT [76, 77]. In the migraine patients, facilitatory responses to the 5 Hz rTMS trains can be observed also at a lower stimulation intensity of 110% of the RMT [78]. Moreover, increased facilitatory responses can be observed at 120% RMT intensity [79]. MEP facilitation during the rTMS trains is cortical in origin and supposed to be mediated by presynaptic glutamate-dependent mechanisms of short-term synaptic plasticity [79–81]. Accordingly, abnormal modulation of the presynaptic mechanisms involved in the control of glutamate release has been supposed to play a role in migraine pathophysiology [78, 79]. Surprisingly, paradoxical inhibitory MEP responses to 5 Hz rTMS trains given at a higher stimulation intensity of 130% RMT have been recorded in migraine patients [78]. This finding has been explained in the light of the homeostatic plasticity of the human cerebral cortex that includes those mechanisms regulating activity of the neural circuits within a physiological useful range [82]. In particular, it has been hypothesized that in migraine, in the interictal period, an altered intracortical excitability characterized by an abnormally increased response to mild stimuli could lead to self-limiting responses to stronger stimuli [78, 83].

Noticeable changes in the MEP response during the rTMS trains have been recorded in episodic migraine patients throughout the different phases of the migraine cycle (i.e., interictal, preictal, ictal, and postictal period) and also in relation to the process of migraine transformation over time [84]. Accordingly, changes in cortical excitability and fluctuations in the threshold for inducing inhibitory homeostatic mechanisms of cortical excitability have been supposed to be relevant for migraine attack recurrence and involved in the process of migraine transformation from episodic to chronic pattern.

9.2.3.2 Studies Targeting the Visual Cortex

The inhibitory aftereffect of 1 Hz rTMS normally observed in the healthy subjects when targeting the visual cortex can be replaced by a paradoxical increase of excitability in the migraine patients [22]. Thus, as for the primary motor cortex, a condition of increased cortical excitability due to either subnormal activation of intracortical GABAergic neurons or hyperresponsivity of the glutamatergic intracortical circuits has been hypothesized. The effects of high-frequency rTMS on the abnormal deficit of habituation to visual- [85, 86] and somatosensory-evoked [87] potentials in migraine have been assessed by some authors. These found that a single session of high-frequency rTMS could restore the normal habituation pattern in migraine patients. In the hypothesis that the lack of habituation to repeated stimuli could be consequence of a reduced interictal thalamocortical drive, it was supposed that high-frequency rTMS could normalize the habituation pattern by increasing the cortical activation level. However, this explanation did not consider the possibility that paradoxical inhibitory effects to high-frequency rTMS could occur when stimulating the visual cortex of migraine patients as observed when targeting the motor cortex. Indeed, in a such condition, the effects of rTMS on the habituation to sensory stimuli could be reinterpreted in the context of cortical homeostatic plasticity. According to the Bienenstock–Cooper–Munro (BCM) model [88] in a condition of enhanced cortical activity, the threshold for induction of LTP and LTD can increase and decrease, respectively. Thus, it is conceivable that if, on the one hand, high-frequency rTMS induces LTP in the healthy subjects, on the other hand, in a condition of increased cortical excitability like that supposed to be present in migraine, it could allow the intracortical circuits to reach the threshold for LTD induction. In this view, the induction of inhibitory regulatory mechanisms counteracting cortical hyperresponsivity could restore a normal habituation pattern. Such a phenomenon has been documented through an experimental model of visual cortical hyperexcitability induced by light deprivation in the healthy subjects [89]. Moreover, some experimental findings may support the idea that in migraine, an increased level of cortical activity as consequence of hyperresponsivity to different kinds of stimuli could lead to a compensatory decrease in the threshold for inhibitory homeostatic responses in the interictal period. In particular, there are two magnetic resonance spectroscopy (MRS) studies showing signs of increased activity of the visual cortex in migraine that cannot further increase neither after prolonged visual stimulation [90] nor after applying a session of anodal transcranial direct current stimulation (tDCS) [91]. In particular, Siniatchkin et al. [91] showed that anodal tDCS, that was capable to increase the activity of the visual cortex in the healthy subjects, could have a paradoxical inhibitory effect when applied over the visual cortex of migraine patients.