Stimulation of dural vessels excites the SI somatosensory cortex of the cat via a relay in the thalamus

Abstract

Aim

We carried out experiments in cats to determine the thalamo-cortical projection sites of trigeminovascular sensory neurons.

Methods

1) We stimulated the middle meningeal artery (MMA) with C-fibre intensity electrical shocks and made field potential recordings over the somatosensory cortical surface. 2) We then recorded neurons in the ventroposteromedial (VPM) nucleus of the thalamus in search of neurons which could be activated from the skin, MMA and superior sagittal sinus. 3) Finally, we attempted to antidromically activate the neurons found in stage 2 by stimulating the responsive cortical areas revealed in stage 1.

Results

VPM neurons received trigeminovascular input, input from the V1 facial skin and could also be activated by electrical stimulation of the somatosensory cortex. VPM neurons activated from the cortex responded with short and invariant latencies (6.7 ± 7.7 msec mean and SD). They could follow high rates of stimulation and sometimes showed collision with orthodromic action potentials.

Conclusions

We conclude that somatosensory (SI) cortical stimulation excites trigeminovascular VPM neurons antidromically. In consequence, these VPM neurons project to the somatosensory cortex. These findings may help to explain the ability of migraineurs with headache in the trigeminal distribution to localise their pain to a particular region in this distribution.

Keywords

Migraine pain somatosensory cortex thalamus trigeminovascular

Introduction

The pain of migraine is qualitatively different from other pains (1), is usually intense (2) and has a strong affective component (3). It has the generalised quality of visceral pain but is partially localisable by patients. It is sometimes felt in a particular cutaneous sensory field, mostly the first (V1) and second (V2) trigeminal fields. Often it is felt in deeper intracranial regions perceived to underlie the cutaneous fields, e.g. retro-orbitally or deeper. The allodynia of migraine is often constrained to the skin of the face and scalp, but in fields consistent with the perceived location of the pain; however, it can spread more widely, even to the whole body (4). The sources of intracranial pain are mostly the large penetrating arteries, the dura mater and, more especially, the blood vessels embedded in the dura mater (5) – along which most of the sensory fibres course. These observations suggest a degree of convergence between signals from cutaneous receptive fields (RFs) and signals from deeper structures like the dura mater and blood vessels. These structures are probably the perceived generator of migraine pain (6). Such convergence, for both trigeminal and occipital input, has already been demonstrated by our previous work (7). We do not know whether this is sufficient to account for the sensory-discriminative aspects of migraine pain or whether processes at a higher level may be necessary.

Regardless of whether such convergence occurs in the cortex or at a lower level, cortical neurons receive projections from the thalamus, where some degree of discrimination might already exist. The ventrobasal complex – ventroposteromedial (VPM) nucleus; ventroposterolateral (VPL) nucleus (8) – is considered to be part of a ‘lateral pain system’ involved in sensory-discriminative aspects of pain (9). Previous studies (10 –13) extended knowledge of general thalamic mechanisms to cover trigeminovascular sensory processing. Neurons with nociceptive input from the dura were located in the VPM and the medial nucleus of the posterior complex (POm). These studies confirmed the importance of the lateral thalamus in craniovascular nociception.

The role of the cortex in craniovascular pain processing is less clear, although there are many studies on its role in trigeminal sensation generally. Early experimental studies in cats (14,15) demonstrated the somatotopy of some thalamo-cortical pathways. Later evidence showed that nociceptive trigeminal stimuli excited cortical neurons in primary somatosensory (SI) cortex in cats (16) and monkeys (17). Clinical studies (18) also supported the role of the somatosensory cortex in nociceptive trigeminal sensory transmission. Nociceptive VPM neurons, including some that are tooth-pulp specific, can be antidromically activated from SI (19). The major cortical projections of VPM neurons are to SI and the secondary somatosensory cortex (SII) (8). Tracing studies in rats have tentatively identified some thalamo-cortical projections as part of a craniovascular pain-perception pathway (20).Thalamic neurons with craniovascular sensory input were found to project to the somatosensory, visual and associative cortices, with different patterns for projections from VPM versus POm.

Rationale

In this study we used electrophysiological techniques to determine whether trigeminovascular neurons in the lateral thalamus, particularly VPM, projected to somatosensory cortex in the cat. To do this, we first recorded evoked potentials (EPs) on the surface of the somatosensory cortex in response to electrical stimulation of the middle meningeal artery (MMA) and then explored the lateral thalamus to identify neurons which might relay information to the cortical areas which showed maximal EPs. We then attempted to activate these neurons antidromically from the cerebral cortical areas previously identified.

Some of the results of the work reported here have previously been reported in abstract form (21,22).

Methods

Experiments were approved by the University of NSW (UNSW) Animal Care and Ethics Committee and conformed to its guidelines.

Animal preparation

Seventeen male or female cats, mass of 3.0 ± 0.5 kg (mean ± standard deviation (SD)) were used in these experiments. Cats were anaesthetised with an intraperitoneal injection of α-chloralose 80 mg kg⁻¹, paralysed with gallamine triethiodide (20 mg kg⁻¹) and artificially ventilated with 30% oxygen in air with a Palmer respiratory pump. Halothane (1%–1.5%) was added to the respiratory circuit via a Fluotech^© vaporiser for the duration of skin surgery, muscle reflection and drilling of bone. A local anaesthetic (lignocaine) was used routinely on skin incisions, in the region of femoral vessel cannulation, in the ear canal before mounting in the stereotaxic frame and on the adjacent scalp muscles after exposure of the cortex. Fluid replacement during surgery and the experiment was made by routine intermittent short infusions of 10 ml min⁻¹ normal saline or a more prolonged slow drip of 0.5 ml min⁻¹ and was always given if mean arterial blood pressure (BP) fell below 80 mmHg. End expiratory CO₂ was maintained at 3.5% to 4%. Temperature was monitored by a rectal thermistor and maintained at 37℃ to 38℃ by a servo-controlled heating blanket. Throughout the surgery and experimental protocol the depth of anaesthesia was assessed regularly by testing for sympathetic responses to noxious stimulation. Resting pupillary slit width was examined hourly. A slit width of greater than 1.5 mm was regarded as indicative of possible low anaesthesia level. The response of the pupils to paw pinch and to electrical stimulation of the superior sagittal sinus (SSS) was checked at the same time. Any widening of the pupils to these stimuli was regarded as indicating the need for supplementary anaesthesia. Resting, and stimulus-induced rises in, BP were tested simultaneously with the tests on pupillary diameter. A resting mean BP of greater than 120 mmHg was regarded as indicating a lowered level of anaesthesia. Any rise in response to paw pinch or electrical stimulation of the SSS was regarded as indicative of a decreased anaesthesia level and supplementary anaesthetic was administered as a bolus dose of 20 mg kg⁻¹ chloralose, administered via the femoral venous catheter.

Cats were mounted in a David Kopf model 1230 stereotaxic frame. A midline 25 mm diameter circular craniotomy was made to expose the SSS. This craniotomy was also used for access to thalamic recording sites. The SSS was subsequently isolated from the underlying cortex by making incisions in the dura on either side of, and parallel to, the SSS. The dura was lifted up from the cortex, elevating the SSS. Another incision was made through the falx, and a sheet of polyethylene was passed through this incision to prevent current spread to the cortex. The SSS was draped over a pair of bipolar hook electrodes, insulated except for the hook area in contact with the SSS. A well was constructed around the craniotomy with a polyethylene cylinder and dental acrylic and filled with paraffin oil to prevent the sinus from drying out and to minimise current spread. A smaller craniotomy was made over the temporoparietal region on the left side of the skull to expose the MMA. Through two cuts in the dura, made 10 mm away from the vessel, a sheet of polyethylene was slipped between the dura and cortex. Miniature chlorided silver ball electrodes were carefully positioned on either side of the artery and transverse to its course. The exposed dura and vessel were then bathed in mineral oil.

To allow access to the somatosensory cortex, a craniotomy was made on the side contralateral to the stimulated MMA to expose an area bounded caudally by the ansate sulcus, anteriorly by the cruciate sulcus and ventrolaterally by a line 8 mm from the length of the coronal sulcus. The dura was removed and a photograph taken, or a drawing made, of the cortex for marking the sites of subsequent cortical recording or stimulation.

Stimulation procedures

The SSS or MMA was stimulated with stimulus-isolated, supramaximal single shocks (40–150 V, 0.4–1.5 mA) of 250 μs duration, every 3 to 5 seconds. For evoked field potential recordings on the cortical surface, only the MMA was stimulated preferentially to minimise stimulus artefact.

Cortical field potential recordings

Potential craniovascular somatosensory projection areas were identified by recording surface potentials evoked by MMA stimulation on the SI and SII somatosensory areas, in particular the ‘face area’ on the coronal gyrus (23). In 11 cats, recordings were made from up to 48 cortical sites. Field potentials were acquired from a chlorided silver ball electrode, voltage-referenced to the stereotaxic frame ground. The electrode was placed gently on the cortex and time allowed for any consequential effects on the cortex beneath the electrode to stabilise. The signals were passed to an amplifier of our own design (bandwidth DC to 50 KHz, no line-frequency notch filter), thence to the A/D converter of a microcomputer based on a Z-80 processor and a program written in FORTRAN and Z-80 Assembler language. Potentials were sampled at 10 KHz for 100 msec post-stimulus to produce time voltage signals which were averaged over 50 successive stimuli for each recording site. The resultant averaged signal was plotted immediately on an X-Y plotter and saved on disk for later analysis. Each stimulus site was marked on a drawing or Polaroid photograph of the cortex and the site which produced maximal evoked field potentials was highlighted on the photograph or drawing.

Thalamic neuronal recording

Neuronal recordings were made from single-barrelled glass micropipettes (impedance 10–30 MΩ) or glass-insulated tungsten microelectrodes (24) (resistance 100 kΩ). Electrodes were mounted on the Inchworm^© motor of a Burleigh 6800 piezoelectric microdrive attached to a Narishige heavy-duty stereotaxic carrier in turn attached to the stereotaxic frame. The thalamus, contralateral to the stimulated MMA and ipsilateral to the cortical hemisphere of interest, was explored between 6 and 10 mm anterior to the interauricular plane and 4 to 7 mm lateral to the midline to include the lateral thalamus. Electrodes were lowered by hand to coordinate H = +5 mm; thence with the microdrive in 5 µ steps to as far as H = −5 mm. Neurons with craniovascular input in the lateral thalamus, including the VPM nucleus and its shell region, were sought and their properties characterised as we have previously performed (11,13).

Signals were fed to a Dagan model 2300 intracellular amplifier (pipette electrode) or Neurolog NL100 preamplifier (tungsten electrode) and thence via further amplifiers (bandwidth 300 Hz to 10 kHz, plus a 50 Hz line frequency notch filter) to a window discriminator and to the A/D convertor of the microcomputer. Post-stimulus time histograms (PSTHs) were constructed in real time and saved on disk for later processing. Software, written in Fortran and Assembler languages (25), allowed us to determine the shapes of action potentials initiated spontaneously, orthodromically or antidromically and thus to confirm that the same unit was activated on each occasion.

Cutaneous RFs

Neurons were tested for the presence of an RF using mechanical or electrical stimulation, or both. RFs were classified according to established criteria (26) as nociceptor specific (NS: responding only to pinch with toothed forceps or deep pressure with a surgical probe), wide dynamic range (WDR: responding to pinch or deep pressure and light touch or brush with a small surgical probe) or low-threshold mechanoreceptors (LTM: responding only to brush or light touch). Those responding only to electrical stimulation were classified as ‘electrical’. We searched also for ‘tap’-sensitive neurons, as defined by Davis and Dostrovsky (10). The space-weighted centre (centroid) of each RF thus determined was stimulated electrically with a pair of stimulus-isolated needle electrodes which delivered supramaximal shocks (10 to 120 V, 250 µsec duration, 0.3 Hz).

Mapping of the extent and degree of cortical EPs

An interactive graphics program, written in FORTRAN and Assembler languages, was used to select and measure absolute values of EP components of interest. Stimulus artefact (recorded with the same amplifier settings after the death of the cat at the end of the experiment) was electronically subtracted from the EPs. This procedure also served to exclude the possibility that the signals observed might have arisen via other pathways or non-neural mechanisms (27). The principal wave of the EP was selected for measurement. Isopotential maps were constructed from measured potential Cartesian X-Y-Z plots with a graphics program (Sygraph^©) as contour maps of voltage (Z-axis) versus the map-projected stereotaxic anterior/posterior (A/P or X-axis) and the lateral (L, or Y-axis) coordinates of the recording electrode. Output was printed to transparencies, which were overlaid, at scale, on a photograph of the relevant area of the cortex in the manner of a map projection (as in Figure 1).

Figure 1.

Upper half: Evoked potential mapping. Averaged EPs (recording bandwidth DC to 30 kHz) from 14 sites on the cortical surface of a cat. The MMA was stimulated with supramaximal square wave pulses. EPs were measured with a silver ball electrode and averaged over 50 successive stimulations. The vertical scale bar on the EPs represents 200 μV, positive upward; the horizontal scale bar represents 100 msec. Electrical stimulation of the MMA produced typical cortical field potentials with maximum amplitudes of the principal components occurring in the so-called ‘face area’ of the SI (shading) area of the cortex (23). Lower half: Isopotential contour map and ‘spot-height’ potentials over cat S1 somatosensory cortex in response to supramaximal stimulation of the contralateral MMA. The diameters of the spots are proportional to the potential of the major component of the EPs. Contour interval is 10 micro volt. The X-Y Cartesian coordinate scales are in arbitrary units proportional to the A/P and L coordinates of the cortical surface according to the Berman and Jones atlas (29). The major centre of activity lies in a band running along the outer convexity of the coronal gyrus. Attempts to stimulate thalamic neurons antidromically focussed on this area. The main structures visible in the photograph are, from left to right, the posterior sigmoid gyrus, the ansate sulcus and the anterior suprasylvian gyrus.

Identification of thalamo-cortical antidromic activation of thalamic neurons

Thalamic (VPM) neurons which responded to dura mater stimulation were investigated for possible projections to primary SI, using antidromic stimulation techniques. The site on the exposed cortex, ipsilateral to the thalamic recording site, and which had the largest field potential evoked by contralateral MMA stimulation, was stimulated through surface bipolar chlorided silver ball stimulating electrodes with an interpolar distance of 2 mm. Stimulus duration was 50 µsec and the intensity was current limited to 5 mA by a Grass CCUIA constant current unit. Stimuli were either initiated under timed program control or, for collision testing, by a trigger signal derived from spontaneous or EPs of the neuron. Thalamic neurons were identified as antidromically activated – i.e. from the axon terminal to the neuron cell body (14,28) only if four of the following five criteria were met:

short (<8 ms) latency;

invariant latency (i.e. single-peaked and narrow post-stimulus histogram);

stability of threshold;

ability to follow repetitive stimulation of at least 50 Hz;

successful collision of antidromic impulses occurring spontaneously or evoked orthodromically by electrical stimulation peripherally.

Criteria one through four are particularly indicative of the absence of a synapse between stimulus site and recording site, thus strengthening the case that the neuron has been activated antidromically.

Collision testing was performed using software described previously (25), summarised as follows: single-unit activity of thalamic neurons was monitored for either spontaneous or orthodromically evoked action potentials which triggered the window discriminator. Detection of an action potential, following a software-settable delay, then generated stimulation of the cortex. Responses were recorded as PSTHs and compared with PSTHs recorded in the absence of a prior action potential.

Each criterion was scored as true (‘Y’) or false (‘N’). The occurrence of four or more ‘Y’s out of five for any one neuron will be significant at p < 0.06, and we coded such neurons as ‘antidromically activated’.

Histology

Electrolytic lesions were made by passing 20–25 µA DC cathodal current for 20–30 s through the recording electrode. At the end of each experiment the animal was deeply anaesthetised with barbiturate and perfused with physiological saline and 10% phosphate-buffered formalin. The brain was cut into 50 µm sections on a freezing microtome and the sections stained with cresyl violet. Recording sites were determined by identifying the lesions and by using the microdrive readings to determine the relative positions of other units in each particular run. Drawings of the histological sections were made by projecting the stained sections with magnification up to 25 times. Stained sections were scanned on an Aperio Scanscope^© model XT scanner at 40 × and 600 dpi, for illustration purposes.

Experimental protocols

Experiments were carried out in the following sequence:

The cortex was explored for the site of the maximum cortical field potential which could be evoked by electrical stimulation of the contralateral MMA.

A thalamic recording electrode was used to explore the thalamus for neurons which responded to MMA stimulation.

For each thalamic neuron so identified, the following neuronal response properties were determined: existence of an RF, its location on the skin and its modality; existence of responses to SSS and MMA electrical stimulation and; measurement of latencies and number of responses to each form of stimulation.

Each neuron was then tested for antidromic activation from the identified cortical region, using the methods described above.

Protocol 2 was then repeated and, for each new responsive neuron identified, Protocols 3 and 4 were repeated.

Results

Cortical EPs

Field potentials produced by stimulation of the contralateral MMA were recorded and mapped in 11 cats and the location of the maximum peak-to-peak amplitude was identified. Sample EPs from mapping in one cat are shown in the upper half of Figure 1 and the resultant isopotential chart in the lower half. For the most part, maximum-amplitude EPs occurred in the SI somatosensory cortex in the regions designated by Felleman et al. (23) as being devoted to facial, orbital and cranial RFs. These correspond to the V1 and V2 divisions of the trigeminal nerve. The average absolute amplitude of the principal component of the maximal EP was 291 ± 207 µvolt, the ‘trough’ occurred at 16 ± 15 msec and the subsequent ‘peak’ at 47 ± 24 msec. These correspond to an A-δ conduction velocity from the periphery to the cortex and are about 20 msec longer than the mean conduction time (as measured from single neuron latencies) to the VPM in this report and our previous report (11).

Thalamic neurons responsive to stimulation of MMA, SSS and craniofacial skin

A total of 37 VPM neurons which responded to electrical stimulation of the MMA or SSS were admitted to the study.

Location

Neurons were located from 3.6 mm above atlas zero to 2.5 mm below it, with a mean of + 0.4 mm. A photomicrograph of the thalamus, including the VPM, at the A/P + 7.5 mm level, marked to show the site of a neuron lesioned during an experiment, is shown in panel (c) of Figure 2.

Figure 2.

Summary of results from a single VPM neuron in one cat, showing: (a) Receptive field of a neuron, with the ‘sensory centroid’ (dark) and surround (hatched). (b) Oscilloscope traces showing action potentials produced by stimulation of the MMA. (c) Histological section of the thalamus, showing three electrolytic lesion sites (black arrows, the recording site is marked by the larger arrow) made at the recording site at A/P level 7.5, corresponding to plates 87/88 of the atlas of Berman and Jones (29). (d) Drawing of the thalamus showing the location of the lesions sites (stars) and the recorded neuron (larger star). (e) PSTHs showing discharges of the same VPM neuron in response to various stimuli as follows: (e1) SSS; (e2) MMA; (e3) skin (RF); (e4) SI somatosensory cortex at 10 msec after spontaneous discharges of the neuron, showing the non-occurrence of collision by the presence of evoked responses; (e5) S1 somatosensory cortex at 1 msec after spontaneous discharges of the neuron, showing the occurrence of collision by the absence of evoked responses. The green bars in e4 and e5 indicate the position of the stimulus artefact resulting from cortical stimulation. They are not neuronal responses.

Cutaneous receptive fields

Twenty-six neurons had an RF detectable by mechanical (N = 18) or electrical (N = 8) probing of the facial skin. These RFs varied in size and sometimes covered more than one trigeminal division; 16 were exclusively in V1 and seven exclusively in V2. Three other neurons were classified as having a V1/V2 RF. Of the 26 neurons, 13 were responsive to mechanical stimulation of the RF in a manner which was readily classifiable as LTM (eight), WDR (three) and NS (one); one neuron was classified as being ‘tap’ sensitive. Eight neurons had RFs that were detectable only by exploratory electrical stimulation. A further five neurons had response modalities that were not easily classifiable. The most responsive spot on the RF was often at the centroid of the RF. A sample cutaneous RF is shown in Figure 2, panel (a).

Figure 3 shows (a) the centroids (red dots) of several RFs; these spots became the stimulus sites for antidromicity testing. These are the locations at which maximum responses (area of the PSTH histogram = number of discharges) could be evoked by dural stimulation. The star represents the RF of the neuron of Figure 2; (b) a mapping of the positions of cortical stimulus sites or maximum MMA-evoked field potentials across the cortex and; (c) a sub-set of eight neurons against a somatotopic map of the cat SI cortex as designated by Felleman et al. (23).

Figure 3.

Somatotopic map for a representative selection of neurons. (a) Stimulus sites on the craniofacial skin, generally at the centroid of a facial RF, from which evoked responses of cortically projecting neurons in the VPM nucleus of the thalamus could be produced. Neurons responsive to such stimuli also responded to electrical stimulation of the MMA and/or the SSS. The site marked with a star is the stimulus site for testing responses of the neuron for which the full extent of the RF is shown in Figure 2. The stimulus sites marked with boxes represent the stimulus sites of two neurons which demonstrated collision of orthodromic and antidromic action potentials. The latter two stimulus sites are mapped to the sites on the cortex where maximum evoked field potentials were obtained (green arrows right upper and bottom). (b) Sites on the surface of the somatosensory cortex (including inset miniature in the left bottom corner of the panel) where maximum-amplitude EPs were measured according to the methods in the text and Figure 1. Most sites were in S1 (green shading). These were the sites which were stimulated electrically to detect possible thalamo-cortical projections of the VPM neurons in the study. (c) This panel shows a selection of stimulus sites from the upper right panel which could be mapped to the SI somatotopic map of Felleman et al. (23). This mapping (especially that of the two sites boxed in red) suggests that RFs tended to be peri-orbital and that the cortical terminations of projections from the VPM nucleus were in the appropriate region of the S1 somatosensory cortex.

Responses to stimulation

Of the 37 neurons admitted to the study, all responded to SSS stimulation, 34 responded to MMA stimulation and 26 responded to stimulation of the face or scalp. Twenty-six neurons responded to all three modes of stimulation. Most neurons responded to stimulation of dural vessels or skin with latencies in the 10 to 20 msec range. Such latencies were about 5 to 10 msec longer than those we have previously observed in the trigeminal nucleus of cats (11) and the same as those which we observed in the thalamus (11). Latencies were mostly equivalent to A-δ conduction velocities in trigeminal sensory fibres although one neuron had a latency reflecting C-fibre conduction velocity peripherally. Figure 4 (left) shows oscilloscope traces of action potentials generated in one neuron by different modes of stimulation and; (right) compares these action potentials with one another through an overlay generated by the averaging software, thus indicating that all action potentials originated from the same neuron. Neuronal latencies are summarised in Table 1.

Figure 4.

Left: Photographs of oscilloscope traces showing action potentials evoked in a cat VPM neuron by electrical stimulation of (A) MMA, (B) SSS, (C) cerebral cortex and (D) craniofacial skin. Vertical bar = 200 μV; horizontal bar = 20 msec. Right: Averaged action potentials from another neuron, showing the similarity of action potential shapes for the four modalities, confirming that all arose from the same neurons. The action potential shape of the cortically evoked action potential is distorted by the large stimulus artefact and the presence of the axon hillock.

Table 1.

Stimulus-response latencies of thalamic neurons in this study.

Latencies (msec)
MMA stimulus			SSS stimulus			RF stimulus
N = 34			N = 36			N = 26
Leading edge	Peak	# Response	Leading edge	Peak	# Response	Leading edge	Peak	# Response
Min	2.4	3.2	0.5	5.0	8.0	0.3	0.4	3.8	0.3
Max	31.2	64.8	16.3	30.4	64.0	8.4	44.0	72.0	9.7
Mean all neurons	16.1	21.1	2.6	14.6	20.3	2.6	14.9	19.0	2.5
SD	7.5	10.8	2.7	6.9	10.6	2.1	8.6	11.9	1.9

Latencies (in msec) from stimulus to the evoked action potentials of responsive neurons in the thalamus. Neurons discharged with predominantly A-δ latencies to stimulation of the MMA, the SSS and the cutaneous RF.

The shortest latency (left-hand edge of the PSTH) and the time of the peak in the PSTH are both tabulated, as is the mean number of responses of the neurons to stimulus. Minimum, maximum, mean and standard deviations for the two latency measures are shown. The number of responses is the integrated area under the PSTH.

MMA: middle meningeal artery; SSS: superior sagittal sinus; RF: receptive fields; SD: standard deviation; PSTHs: post-stimulus time histograms.

Tests for direct thalamo-cortical projections

Of the 37 neurons in VPM, 33 could be activated by stimulation of the somatosensory cortex. A total of 12 neurons fulfilled at least four of the five criteria which we set for considering a neuron as antidromically activated from the cortex. Twenty-one of the 33 neurons were subjected to collision testing, but only three were positively identified as demonstrating the occurrence of collisions. These results are summarised in Table 2. Figure 5 (left) shows two oscilloscope traces which illustrate the collision of orthodromic and antidromic spikes and (right) the histograms resulting from 50 repetitions of this test. Figure 6 shows PSTHs resulting from repetitions of these four modes of stimulation.

Figure 5.

Left panels: Photographs of oscilloscope traces showing: Upper: Response of a VPM neuron to stimulation of the somatosensory cortex. The neuron responded to this stimulus with a latency of 2 msec. (A) Cortical stimulation artefact; R1: Response to stimulus. Lower: Failure of the neuron to respond to cortical stimulation triggered by the occurrence of an action potential of the VPM neuron evoked by stimulation of the facial RF, indicating the occurrence of collision and hence the antidromic nature of the cortical stimulus. R2: response of neurons to RF stimulation; (A) Artefact from cortical stimulation. No action potential follows this. Right panels: Post-stimulus time histograms constructed from the responses of a (different) neuron. Upper: Response (arrowed) of the neuron to stimulation in the SI somatosensory cortex, 10 msec after the occurrence of an action potential evoked by RF stimulus. Lower: lack of responses of the neuron to stimulation in the SI somatosensory cortex, 1 msec after the occurrence of an action potential evoked by RF stimulus, indicating the collision of the antidromically generated action potential with the orthodromic action potential. The green histogram bars represent stimulus artefact.

Figure 6.

Responses of a VPM neuron to: (A) SSS stimulation, (B) MMA stimulation, (C) RF stimulation, (D) cortex (Cx) stimulation. This neuron responded with A-δ latencies (20 to 40 msec) to stimulation of the skin or dura mater vessels and with a latency of 6 msec to stimulation of the SI somatosensory cortex.

Table 2.

Summary of the results of testing 37 ventroposteromedial (VPM) neurons for their direct projections to the cortex.

Total stimulated	Total responding	Short latency?	Invariant latency?	Follow high-freq. stim?	Collision?	Meet ≥ 4 criteria?
37	33	22	19	14	3	12

Summary of the tests of the response properties of thalamic neurons that could be activated by electrical stimulation of the cerebral cortex. All neurons responded to stimulation of dural blood vessels and the skin in previous testing. Neurons were tested for (1) activation from the cortex, (2) the latency length for this activation (it should be short), (3) the consistency of this latency (it should be consistent), (4) their ability to follow high-frequency stimulation (they should) and (5) the occurrence of collision between a cortically evoked action potential and a prior action potential (collision confirms antidromic activation). Thirty-three neurons responded and 12 of them fulfilled at least four of these criteria and were held to have been antidromically activated from the cortex. These neurons were accordingly considered to be cortically projecting neurons.

Discussion

The aim of these experiments was to find trigeminovascular thalamic neurons which projected to the sensory cortex and to determine which areas of the cortex these were. It was not possible to record single neurons simultaneously in both loci, so the initial search for potential cortical projection areas was carried out by stimulating the MMA electrically and recording field potentials over an extensive area of the cortex. The cortical region found to be the most responsive to such stimulation thence became the site of antidromic activation of identified thalamic neurons in later stages of the experiment. The MMA was chosen in preference for the probing stimulus site because it was further away from the recording site and hence easier to limit current spread – and hence artefact – from this smaller vessel.

Our experiments were carried out in cats, which have a gyrencephalic cortex more similar to that of humans than is the lissencephalic cortex of rats. The gyrencephalic cortex of humans is almost certainly involved in human migraine, so a species with such a cortex makes for a better model than a species with a lissencephalic cortex.

In this study we asked three questions: Does noxious stimulation of the dura produce effects on the cortex? Do these cortical areas make somatotopic sense? and Could such effects be relayed via a thalamic projection? For a significant number of the neurons tested, the answer was ‘yes’.

We used an EP mapping technique used by others (14,30) to reveal SI areas receiving thalamic projections. It is not obvious from which cortical layer(s) the potentials arose, but it was probably from deeper pyramidal neurons, as concluded by Darian-Smith (14).

The EP mapping technique showed that field potentials were maximal in the ‘face’, ‘orbital’ and ‘cranial’ areas of SI (23). Figure 3 illustrates how the thalamic somatotopic map projected to make a map of the trigeminovascular SI.

Penfield and Boldrey were first to ‘map’ areas of the human sensory cortex to anatomical structures (31). Human studies (32) were complemented by animal studies in rat (33), cat (23) and two monkey species (34 –36). Representation of exclusively facial cutaneous (35 –38) and oral structures (32) are less well studied. There is little agreement on the somatotopy of facial sensation (34). The representation of the dura mater is even less well studied (20 –22). We found that cortically projecting VPM trigeminovascular thalamic neurons map to the periorbital trigeminal distribution of the face as defined by Felleman et al. (23). This is consistent with the distribution found by Noseda et al. in rats (20).

In this study, we used electrical stimulation of C-fibre intensity in order to activate all possible fibres innervating the dura and the skin. It has been emphasised (39) that the pain of migraine must be relayed by fibre systems which have appropriate characteristics such as C-fibre latencies and NS receptive fields. However, as with several other studies (including our own (11)) in a number of species (20,30,40), we found few thalamic neurons and no SI cortical areas which responded with C-fibre latencies – even where the thalamic neurons had NS receptive fields. Ivanusic et al. (30) also recorded cortical field potentials evoked by C-fibre intensity peripheral stimuli and also found only A-δ latency responses. In the study by Noseda et al. (20) all four of the neurons activated by noxious trigeminovascular stimulation and projecting to somatosensory cortex also responded with A-δ latencies only. Similarly, in the report by Shields and Goadsby (40), only one VPM neuron which responded to trigeminovascular stimulation had an NS receptive field, while 11 others had WDR fields. The reason why there are so few C-fibre latencies and NS receptive fields found in these studies, and in our study, is unknown.

There are many studies on structural and functional imaging in headache (41 –45). Human cortical regions which appear to have associations with headache include the anterior cingulate (41,46,47), insula (48,49), prefrontal cortex (50) and the temporal lobes (51,52). The somatosensory cortex is missing from this list. These areas, and several sub-cortical areas, were dubbed the ‘pain matrix’ (53), a description recently used to explain migraine headache (54). It is not universally accepted that the ‘pain matrix’ implies pain somatotopy (55,56). Migraineurs can often ‘localise’ the pain of their headache, especially when it presents as extracranial pain, but also when it presents as an intracranial pain. The former is easily explained on the basis of convergence of fibres at the first synaptic level. The latter is harder to explain and might more reasonably be explained by thalamic or cortical somatosensory maps. It may be a leap of faith to search for a neuronal population specific for craniovascular nociception. The finding by Noseda et al. (20) that cortical projections of craniovascular thalamic neurons in POm are diffuse and widespread reinforces this notion. There does not appear to be such a thing as a viscerotopic map (56), but our results are not inconsistent with some kind of mapping of noxious input signals to cortical areas which make ‘somatotopic sense’.

We found that electrical stimulation of the cortex excited trigeminovascular thalamic neurons by what appeared to be antidromic activation. We applied five tests to determine whether trigeminovascular VPM neurons were activated antidromically. Neurons which fulfilled four of the five criteria were accepted as being activated antidromically. Of the five tests, the collision test is the most definitive (14) but the hardest to carry out and interpret, principally because of the occurrence of responses in a narrow time window subject to stimulation artefact caused by current spread. Only three neurons passed this test and we can say that only three neurons were unequivocally activated by an antidromic pathway. However, we can say with 95% confidence that 12 neurons were activated this way.

Nociceptive neurons in SI exist in cat (16,57) and primate (17) cortex. In the cat, nociceptive VPM neurons project to SI and SII while POm neurons project to SII, SIII and SIV as well as SI (57). Nociceptive VPM neurons in SI and SII of monkeys have characteristics which suggest involvement in the sensory-discriminative aspects of pain (17,58) because they have small, contralateral somatotopically arranged RFs. We have described similar nociceptive trigeminovascular thalamic neurons in cat VPM (11,13). The role of the POm in nociception has been somewhat controversial (8,9). However, we identified nociceptive POm neurons in the cat with craniovascular input but their RFs were often broad and involved the ipsilateral, as well as the contralateral, face (11).

Burstein’s group has reported on the cortical projections of dura-responsive VPM and POm neurons in the rat (59). They showed that a small number of dural-/light-sensitive thalamic neurons in posterior thalamus, including POm, projected to layers I-V of SI and other cortical regions. In another study (20), they showed that dura-responsive neurons in VPM projected primarily to trigeminal regions of SI, SII and insular cortex. In contrast, dura-responsive neurons in the POm, lateral posterior (LP) and lateral dorsal (LD) nuclei projected to trigeminal and non-trigeminal regions of SI and SII and many other cortical regions. They concluded that dura-sensitive VPM neurons were involved in the sensory-discriminative aspects of migraine pain, while the more widespread projections of POm and LD/LP neurons to multiple cortical regions including auditory, visual and association cortices could be responsible for the other non-pain neurological aspects of migraine.

In this study, we examined neurons in VPM. This is only one of several nuclei with craniovascular sensory associations (11,13,59 –62). However, it may be an important one for processing sensory-discriminative aspects of migraine headache (20).

The somatotopic map in Figure 3 gives some insight as to why the visceral pain of migraine can be mapped by migraineurs to a trigeminal cutaneous sensory distribution. This ability must be dependent on convergence between cutaneous and intracranial sensory fibres. The carriage of noxious and non-noxious sensory information is already segregated in the periphery (63). This is partly because of a convergence on second-order sensory neurons rather than divergent axon collaterals (64,65). This segregation is carried forward to the thalamus (66,67) and possibly the cortex (68). The principal sensory nucleus and the nucleus caudalis of the trigeminal complex project to chemically distinct regions of the VPM; this differential pattern is carried forward to separate cortical layers (69).

In our experience, it is hard to find exclusively dura mater-specific neurons (i.e. without any cutaneous RF) in the trigeminal nucleus (7,70), the thalamus (11,71) or the cortex (this work). It is easy to find facial, occipital, corneal, tooth pulp or even limb RFs which will activate trigeminovascular neurons. Thus, it is too much to expect that the somatosensory cortex receives inputs carrying exclusively trigeminovascular noxious information.

Not only are pain signals segregated from non-pain signals in the trigeminal system, but trigeminovascular pain and other trigeminal pain signals are segregated from each other as well. This occurs before the first sensory synapse. We have formed the view that this is related to the nature of the presynaptic 5-HT_1B or 5HT_1D receptors on incoming sensory fibres (72 –75).

The cerebral cortex might generate migraine pain (76 –82). A number of regions are implicated – visual, cingulate, somatosensory and orbito-frontal cortices. In the rat somatosensory cortex, most descending fibres to the midbrain and trigeminal nucleus appear to arise from large pyramidal neurons in layer V (83). These may be responsible for generating the EPs which we recorded here. In migraine with aura, visual symptoms are usually followed by headache. Flickering light can also trigger migraine with a short latency (84). These observations favour the visual cortex as a site of origin, but it is unclear whether it is necessary that any particular region of the cortex is linked to the generation of migraine pain. It was recently reported that rat somatosensory cortex is influenced by craniovascular sensory processing at the trigeminal level (85). If this pathway is relevant to migraine it could form part of a feedback loop that could account for a number of the features of migraine.

Conclusions

This electrophysiological study is the first to demonstrate in a non-rodent species that thalamic neurons which receive nociceptive input from the dura mater and its vessels project to anatomically relevant areas of the SI somatosensory cortex. The results provide a basis for explaining why migraineurs often can ‘localise’ the intracranial visceral pain of migraine. They do not give any information as to why migraine pain has unique qualities such as its associated signature symptoms. These are probably due to projections from other thalamic nuclei, such as POm, which are much more extensive (20).

Clinical implications

We conducted the first fully electrophysiological study in a non-rodent species of thalamo-cortical projects of the trigeminovascular system.

Our results reinforce the importance of the lateral thalamus in the relay of nociceptive trigeminovascular sensation.

Our work supports the idea that dural sensation is relayed to ‘logical’ areas of the somatosensory cortex.

This research raises questions about the role of A-δ versus C-fibre peripheral sensory conduction in dural sensation.

Footnotes

Funding

This work was supported by the Australian National Health and Medical Research Council, the Australian Brain Foundation, the J.A. Perini Family Trust and Warren and Cheryl Anderson.

Conflict of interest

None declared.

Acknowledgements

We would like to thank Dr Victor Gordon for help with the experiments, Professor George Paxinos of Neuroscience Research Australia for scanning the histological sections and Craig Hardman for preparing them for scanning.

References

Kelman

. Pain characteristics of the acute migraine attack. Headache 2006; 46: 942–953.

Cady

Freitag

Mathew

. Allodynia-associated symptoms, pain intensity and time to treatment: Predicting treatment response in acute migraine intervention. Headache 2009; 49: 350–363.

Singer

Seymour

O'Doherty

. Empathy for pain involves the affective but not sensory components of pain. Science 2004; 303: 1157–1162.

Burstein

Jakubowski

Garcia-Nicas

. Thalamic sensitization transforms localized pain into widespread allodynia. Ann Neurol 2010; 68: 81–91.

Lance

Goadsby

. Mechanism and management of headache, 7th edn. London: Butterworth Scientific, 2005.

Ray

Wolff

. Experimental studies on headache. Pain-sensitive structures of the head and their significance in headache. Arch Surg 1940; 41: 813–856.

Angus-Leppan

Lambert

Michalicek

. Convergence of occipital nerve and superior sagittal sinus input in the cervical spinal cord of the cat. Cephalalgia 1997; 17: 625–630.

Jones

. The thalamus, New York: Plenum Press, 1985.

Albe-Fessard

Berkley

Kruger

. Diencephalic mechanisms of pain sensation. Brain Res 1985; 356: 217–296.

10.

Davis

Dostrovsky

. Properties of feline thalamic neurons activated by stimulation of the middle meningeal artery and sagittal sinus. Brain Res 1988; 454: 89–100.

11.

Zagami

Lambert

. Stimulation of cranial vessels excites nociceptive neurones in several thalamic nuclei of the cat. Exp Brain Res 1990; 81: 552–566.

12.

Goadsby

Zagami

. Stimulation of the superior sagittal sinus increases metabolic activity and blood flow in certain regions of the brainstem and upper cervical spinal cord of the cat. Brain 1991; 114: 1001–1011.

13.

Zagami

Lambert

. Craniovascular application of capsaicin activates nociceptive thalamic neurones in the cat. Neurosci Lett 1991; 121: 187–190.

14.

Darian-Smith

. Cortical projections of thalamic neurones excited by mechanical stimulation of the face of the cat. J Physiol 1964; 171: 339–369.

15.

Albe-Fessard

Levante

Rokyta

. Cortical projections of cat medial thalamic cells. Int J Neurosci 1971; 1: 327–338.

16.

Matsumoto

Sato

Yahata

. Physiological properties of tooth pulp-driven neurons in the first somatosensory cortex (SI) of the cat. Pain 1987; 31: 249–262.

17.

Chudler

Anton

Dubner

. Responses of nociceptive SI neurons in monkeys and pain sensation in humans elicited by noxious thermal stimulation: Effect of interstimulus interval. J Neurophysiol 1990; 63: 559–569.

18.

Kobari

Meyer

Ichijo

. Hyperperfusion of cerebral cortex, thalamus and basal ganglia during spontaneously occurring migraine headaches. Headache 1989; 29: 282–289.

19.

Yokota

Nishikawa

Koyama

. Tooth pulp input to the shell region of nucleus ventralis posteromedialis of the cat thalamus. J Neurophysiol 1986; 56: 80–98.

20.

Noseda

Jakubowski

Kainz

. Cortical projections of functionally identified thalamic trigeminovascular neurons: Implications for migraine headache and its associated symptoms. J Neurosci 2011; 31: 14204–14217.

21.

Lambert

Zagami

. Cortical projections of thalamic neurons which receive trigeminovascular input. Soc Neurosci Abstr 1995; 21: 112–112.

22.

Zagami

Lambert

. Thalamic neurones with craniovascular input project to somatosensory cortex in the cat. J Clin Neurosci 1997; 4: 412–412.

23.

Felleman

Wall

Cusick

. The representation of the body surface in S-I of cats. J Neurosci 1983; 3: 1648–1669.

24.

Levick

. Another tungsten microelectrode. Med Biol Eng 1972; 10: 510–515.

25.

Lambert

. Software-controlled testing for antidromic activation in neurophysiology. CABIOS 1994; 10: 237–241.

26.

Dostrovsky

Sessle

. Functional properties of neurons in cat trigeminal subnucleus caudalis (medullary dorsal horn). I. Responses to oral-facial noxious and non-noxious stimuli and projections to thalamus and subnucleus caudalis. J Neurophysiol 1981; 45: 173–191.

27.

Lambert

Zagami

. Trigeminovascular sensory signals CAN be modulated by central mechanisms. A response to a Cephalalgia Viewpoint. Cephalalgia 2013; 33: 347–350.

28.

Paintal

. Intramuscular propagation of sensory impulses. J Physiol 1959; 148: 240–251.

29.

Berman

Jones

. The thalamus and basal telencephalon of the cat. A cytoarchitectonic atlas with stereotaxic coordinates, Madison, WI: University of Wisconsin Press, 1982.

30.

Ivanusic

Sahai

Mahns

. The cortical representation of sensory inputs arising from bone. Brain Res 2009; 1269: 47–53.

31.

Penfield

Boldrey

. Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation. Brain 1937; 60: 389–443.

32.

Nguyen

Tran

Hoshiyama

. Face representation in the human primary somatosensory cortex. Neurosci Res 2004; 50: 227–232.

33.

Krieg

. Connections of the cerebral cortex. I. The albino rat. A. Topography of the cortical areas. J Comp Neurol 1946; 84: 221–275.

34.

Flaherty

Graybiel

. Two input systems for body representations in the primate striatal matrix: Experimental evidence in the squirrel monkey. J Neurosci 1993; 13: 1120–1137.

35.

Nelson

Sur

Felleman

. Representations of the body surface in postcentral parietal cortex of Macaca fascicularis. J Comp Neurol 1980; 192: 611–643.

36.

Sur

Garraghty

Bruce

. Somatosensory cortex in macaque monkeys: Laminar differences in receptive field size in areas 3b and 1. Brain Res 1985; 342: 391–395.

37.

Yang

Gallen

Schwartz

. Noninvasive somatosensory homunculus mapping in humans by using a large-array biomagnetometer. Proc Natl Acad Sci U S A 1993; 90: 3098–3102.

38.

Yang

Gallen

Schwartz

. Sensory maps in the human brain. Nature 1994; 368: 592–593.

39.

Zhang

Levy

Kainz

. Activation of central trigeminovascular neurons by cortical spreading depression. Ann Neurol 2011; 69: 855–865.

40.

Shields

Goadsby

. Propranolol modulates trigeminovascular responses in thalamic ventroposteromedial nucleus: A role in migraine? Brain 2005; 128: 86–97.

41.

Sprenger

Goadsby

. What has functional neuroimaging done for primary headache and for the clinical neurologist? J Clin Neurosci 2009; 17: 547–553.

42.

Paul

. What has imaging taught us about migraine? Maturitas 2011; 70: 34–36.

43.

Sánchez del Rio

Alvarez Linera

Aminoff

. Neuroimaging in headache. Handbook of clinical neurology, Amsterdam: Elsevier, 2004, pp. 781–787.

44.

Sánchez del Rio

Alvarez Linera

. Functional neuroimaging of headaches. Lancet Neurol 2004; 3: 645–651.

45.

May

. Morphing voxels: The hype around structural imaging of headache patients. Brain 2009; 132: 1419–1425.

46.

May

Bahra

Buchel

. Hypothalamic activation in cluster headache attacks. Lancet 1998; 352: 275–278.

47.

Schmidt-Wilcke

Gänssbauer

Neuner

. Subtle grey matter changes between migraine patients and healthy controls. Cephalalgia 2008; 28: 1–4.

48.

Afridi

Giffin

Kaube

. A positron emission tomographic study in spontaneous migraine. Arch Neurol 2005; 62: 1270–1275.

49.

Kim

Suh

Seol

. Regional grey matter changes in patients with migraine: A voxel-based morphometry study. Cephalalgia 2008; 28: 598–604.

50.

May

Bahra

Büchel

. PET and MRA findings in cluster headache and MRA in experimental pain. Neurology 2000; 55: 1328–1335.

51.

McAbee

Sagan

Winter

. Olfactory hallucinations during migraine in an adolescent with an MRI temporal lobe lesion. Headache 2000; 40: 592–594.

52.

Rocca

Ceccarelli

Falini

. Brain gray matter changes in migraine patients with T2-visible lesions: A 3–T MRI study. Stroke 2006; 37: 1765–1770.

53.

Talbot

Marrett

Evans

. Multiple representations of pain in the human cerebral cortex. Science 1991; 251: 1355–1358.

54.

May

. Neuroimaging: Visualising the brain in pain. Neurol Sci 2007; 28: S101–S107.

55.

Iannetti

Mouraux

. From the neuromatrix to the pain matrix (and back). Exp Brain Res 2010; 205: 1–12.

56.

Wall

. Independent mechanisms converge on pain. Nat Med 1995; 1: 740–741.

57.

Norio

. Functional difference of tooth pulp-driven neurons in oral and facial areas of the somatosensory cortex (SI) of the cat. Exp Neurol 1984; 85: 437–451.

58.

Waite

PME

Ashwell

KWS

Trigeminal sensory system. In: Paxinos

Mai

(eds). The human nervous system, London: Academic Press, 2004, pp. 1061–1086.

59.

Noseda

Kainz

Jakubowski

. A neural mechanism for exacerbation of headache by light. Nat Neurosci 2010; 13: 239–245.

60.

Noseda

Burstein

. Advances in understanding the mechanisms of migraine-type photophobia. Curr Opin Neurol 2011; 24: 197–202.

61.

Goadsby

Zagami

. Thalamic processing of craniovascular pain in the cat: A 2-deoxyglucose study. Soc Neurosci Abstr 1990; 16: 1144–1144.

62.

Goadsby

Zagami

Lambert

. Neural processing of craniovascular pain: A synthesis of the central structures involved in migraine. Headache 1991; 31: 365–371.

63.

Sessle

Amano

. Convergence of cutaneous, tooth pulp, visceral, neck and muscle afferents onto nociceptive and non-nociceptive neurones in trigeminal subnucleus caudalis medullary dorsal horn and its implications for referred pain. Pain 1986; 27: 219–235.

64.

Borges

Moskowitz

. Do intracranial and extracranial afferents represent divergent axon collaterals? Neurosci Lett 1983; 35: 265–270.

65.

McMahon

Norregaard

Beyerl

. Trigeminal afferents to cerebral arteries and forehead are not divergent axon collaterals in cat. Neurosci Lett 1985; 60: 63–68.

66.

. Response properties of nociceptive and non–nociceptive neurons in the rat's trigeminal subnucleus caudalis (medullary dorsal horn) related to cutaneous and deep craniofacial afferent stimulation and modulation by diffuse noxious inhibitory controls. Pain 1990; 41: 331–345.

67.

Sessle

. Neural mechanisms and pathways in craniofacial pain. Can J Neurol Sci 1999; 26: S7–S11.

68.

DaSilva

Becerra

Makris

. Somatotopic activation in the human trigeminal pain pathway. J Neurosci 2002; 22: 8183–8192.

69.

Rausell

Jones

. Chemically distinct compartments of the thalamic VPM nucleus in monkeys relay principal and spinal trigeminal pathways to different layers of the somatosensory cortex. J Neurosci 1991; 11: 226–237.

70.

Angus-Leppan

Olausson

Boers

. Convergence of afferents from superior sagittal sinus and tooth pulp on cells in the upper cervical spinal cord of the cat. Neurosci Lett 1994; 182: 275–278.

71.

Angus-Leppan

Olausson

Boers

. Convergence of afferents from superior sagittal sinus and tooth pulp on cells in the thalamus of the cat. Cephalalgia 1995; 15: 191–199.

72.

Boers

Donaldson

Zagami

. 5-HT_1A and 5-HT_1B/1D receptors are involved in the modulation of the trigeminovascular system of the cat: A microiontophoretic study. Neuropharmacol 2000; 39: 1833–1847.

73.

Donaldson

Boers

Hoskin

. The role of 5-HT_1B and 5-HT_1D receptors in the selective inhibitory effect of naratriptan on trigeminovascular neurons. Neuropharmacology 2002; 42: 374–385.

74.

Boers

Donaldson

Zagami

. Naratriptan has a selective inhibitory effect on trigeminovascular neurones at central 5-HT_1A and 5-HT_1B/1D receptors in the cat: Implications for migraine therapy. Cephalalgia 2004; 24: 99–109.

75.

Lambert GA, Boers PM and Zagami AS. Triptan-induced sensitization of trigeminovascular sensation. Headache. Epub ahead of print 9 September 2008. DOI: 10.1111/j.1526-4610.2008.01238.x.

76.

Chakravarty

. How triggers trigger acute migraine attacks: A hypothesis. Med Hypotheses 2009; 74: 750–753.

77.

Moulton

Becerra

Maleki

. Painful heat reveals hyperexcitability of the temporal pole in interictal and ictal migraine states. Cereb Cortex 2010; 21: 435–448.

78.

Stankewitz

May

. The phenomenon of changes in cortical excitability in migraine is not migraine-specific – A unifying thesis. Pain 2009; 145: 14–17.

79.

Rogawski

. Common pathophysiologic mechanisms in migraine and epilepsy. Arch Neurol 2008; 65: 709–714.

80.

DaSilva

Granziera

Snyder

. Thickening in the somatosensory cortex of patients with migraine. Neurology 2007; 69: 1990–1995.

81.

Coppola

Pierelli

Schoenen

. Is the cerebral cortex hyperexcitable or hyperresponsive in migraine? Cephalalgia 2007; 27: 1427–1439.

82.

Lambert

Zagami

. The mode of action of migraine triggers: A hypothesis. Headache 2009; 49: 253–275.

83.

Wise

Jones

. Cells of origin and terminal distribution of descending projections of the rat somatic sensory cortex. J Comp Neurol 1977; 175: 129–157.

84.

Harle

Shepherd

Evans

. Visual stimuli are common triggers of migraine and are associated with pattern glare. Headache 2006; 46: 1431–1440.

85.

Noseda

Constandil

Bourgeais

. Changes of meningeal excitability mediated by corticotrigeminal networks: A link for the endogenous modulation of migraine pain. J Neurosci 2010; 30: 14420–14429.