Laser-evoked brain potentials (LEPs) and laser-evoked magnetic fields (LEFs)
1. Laser-evoked potentials elicited by the co-ctivation of Aδ and C fiber nociceptors
Brief thermal stimuli applied to the skin elicit a large negative-positive potential referred to as the N2 and P2 waves or N2-P2 complex. Given the large amplitude of these deflections, the N2 and P2 waves are often visible in single trials, and can be identified and characterized reliably using only a few repeated stimuli (typically, 20-30 stimuli). The scalp topographies of both the N2 and P2 waves are maximal at the vertex, and are largely independent of the location of the stimulus. The N2 and P2 waves are preceded by an earlier negative deflection, often referred to as the N1 wave. At the scalp vertex, the N1 wave often appears embedded in the ascending shoulder of the N2 wave, and is thus difficult to isolate. The scalp topography of the N1 wave is clearly lateralized, appearing maximal over central and temporal regions contralateral to the stimulated side. Xu et al. (1995) suggested that the scalp topography of the N1 wave may be further dependent on the stimulation site. Indeed, they described a contralateral temporal topography when stimulating the hand, and a more central topography when stimulating the foot. However, such a somatotopic shift in N1 scalp topography was not found in other studies examining laser-evoked potentials elicited by the stimulation of different body districts (Tarkka and Treede 1993, Spiegel et al. 1996, Valeriani et al. 1996).
The latency of the N1, N2 and P2 waves of laser-evoked potentials is largely dependent on the peripheral conductance distance ($). The obtained latencies are compatible with the conduction velocity of Aδ fibres, and are incompatible with the conduction velocity of C fibres.
Several studies have attempted to estimate the conduction velocity of the spino-thalamic tracts conveying the afferent volley leading to laser-evoked potentials. For example, by comparing the latencies of laser-evoked potentials elicited by stimuli applied to the skin overlying vertebral spinous processes at various levels (such as to minimize peripheral conduction distance), Cruccu et al. ($) estimated that the conduction velocity of the spino-thalamic tracts generating the P2 wave was 11.2 m/s. Using the same method, Valeriani et al. ($) proposed that the conduction velocity of the spino-thalamic tracts generating the N1 wave was significantly slower (9 m/s) than the conduction velocity of the spino-thalamic tracts generating the later P2 wave (13 m/s), and concluded that both waves reflect nociceptive input conveyed within distinct spino-thalamic pathways. However, the observed differences in spino-thalamic conduction velocities could also be explained by the possibility that some event-related potentials reflect neuronal processes that are triggered by the first arriving afferent inputs (i.e. the afferent inputs conveyed by the fastest fibres) while others reflect neuronal processes that are triggered by the bulk of the arriving afferent input.
A number of studies have applied source analysis methods to laser-evoked brain potentials ($). Albeit some exceptions (Bentley et al. 2001; Iannetti et al. 2003), most of these studies have suggested that bilateral opercular regions contribute largely to the observed LEP waveforms. These opercular sources have been hypothesized to reflect neural activity originating from the secondary somatosensory cortex (S2) and, possibly, from the deeper insular cortex ($). Most studies have identified the contralateral opercular response as the earliest response following laser stimulation, coinciding with the latency of the N1 wave. However, the suggested time courses of opercular activity suggest that they also contribute to the later N2 and P2 waves. In addition to bilateral opercular sources, source-analysis studies have repeatedly proposed that laser-evoked potentials also reflect activity originating from the anterior cingulate cortex, in particular, the midportion corresponding to Brodman’s area 24. In most cases, the latency of this additional source of activity suggests that it contributes mostly to the later N2 and P2 waves. Whether or not the primary somatosensory cortex contributes to the LEP waveform remains a matter of debate. Indeed, several source-analysis studies have shown that the bulk of the LEP waveform can be satisfactorily explained without assigning sources in the contralateral primary somatosensory cortex.
2. Laser-evoked potentials elicited by the selective activation of C fiber nociceptors
A number of studies have examined the laser-evoked brain potentials elicited by the selective activation of C nociceptors, using various methods to avoid the concomitant activation of Aδ nociceptors, or block the transmission of Aδ fibre input at peripheral level ($).
A first method takes advantage of the fact that unmyelinated C fibres are more resistant to a focal nerve compression than myelinated nerve fibres, including small-diameter Aδ fibres. Using this differential property, several studies have applied pressure to the superficial branch of the radial nerve to induce a selective blockade of myelinated nerve fibres ($). These studies have shown that after a various delay (usually, 30-60 minutes), brief laser stimuli above the threshold of both Aδ and C nociceptors applied to the skin innervated by the nerve no longer elicit a sensation of first pain, but still elicit a sensation of second pain. Furthermore, the Aδ fibre-related laser-evoked potentials disappear, and, more surprisingly, ultra-late laser-evoked potentials appear within a latency range compatible with the conduction of C fibres, i.e. approximately one second after stimulus onset.
A second method takes advantage of the difference in thermal activation threshold between heat-sensitive Aδ and C fibre receptorsThe thermal activation threshold of type II Aδ-fibre mechano-heat nociceptors (AMH Type II), which are thought to underlie laser-evoked potentials related to the activation of Aδ nociceptors, lies in the range of 46°C ($). In comparison, C fibre mechano-heat nociceptors have a significantly lower thermal activation threshold, lying in the range of 41°C (Treede 1995). Furthermore, some heat-sensitive C fibre receptors, referred to as C warm receptors, can respond to increases in skin temperatures of less than 1°C above baseline skin temperature (Lamotte and Campbell, 1978). Exploiting these differences, Magerl et al. ($) devised an ingenious protocol based on a feedback controlled CO2 laser stimulator. At a base temperature of about 33°C, the skin was exposed to two successive heat ramps at 5-s interval. A first heat ramp of 50°C/s lasting 150 ms brought the skin temperature to 40°C, allowing the activation of heat-sensitive C nociceptors. Skin temperature was kept constant at 40°C for an additional 5 seconds, and, then, the skin was briefly exposed to a second heat ramp bringing skin temperature to 48°C, allowing this time the activation of Aδ nociceptors. The first heat ramp elicited an ultra-late laser-evoked potential whose latency was compatible with the conduction velocity of C fibres, while the second heat ramp elicited a more precocious response, compatible with the conduction velocity of Aδ fibres.
A third method takes advantage of the fact that certain pathologies can lead to a selective impairment of myelinated nerve fibres. For example, Lankers et al. ($) observed ultra-late laser-evoked brain potentials in a patient with hereditary motor and sensory neuropathy type I, indicating the preservation of unmyelinated C fibre function. Similar anecdotal observations have been made in some painful peripheral neuropathies characterized by a selective loss of Aδ fibre function ($).
Finally, one can exploit the fact that in the skin, the density of C fibre free nerve endings is greater than the density of Aδ fibre free nerve endings. Indeed, depending on the species and methods for quantification, the density distribution of C fibre free nerve endings is about 2-8 /mm2, while the density distribution of Aδ fibre free nerve endings is thought to be <1 /mm2 (Ochoa and Mair 1969, Schmidt et al. 1994). Therefore, if a laser stimulus is applied using a very small stimulus surface area (e.g. 0.15 mm2; Bragard et al. 1996), there is a high probability that the beam will heat C fibre free nerve endings without concomitantly heating Aδ fibre free nerve endings. As compared to other methods, the approach has the advantage of not being invasive, and not requiring to lower the energy density of the stimulus. Several studies (Bragard 1996, Opsommer 1999, 2001, 2001, Qiu et al. 2002, Opommer et al 2003) have successfully used this approach to record ultra-late laser evoked potentials.
Whatever the method used, all of these studies have confirmed that when the activation of Aδ fibre nociceptors is avoided, or when the peripheral Aδ fibre afferent volley is blocked at peripheral level, the selective activation of C fibres elicits an ultra-late negative positive complex whose latency (750-1150 ms after stimulation of the hand) is compatible with the conduction velocity of unmyelinated C fibres. The morphology and scalp topography of these ultra-late responses very much resembles the morphology and scalp topography of the laser-evoked potential elicited by Aδ fibres. The most prominent component of the response also consists of a negative-positive complex, maximal at the vertex, and often referred to as ultra-late N2-P2 ($). Valeriani et al. (2002) reported that, in addition to eliciting ultra-late N2 and P2 components, the selective activation of C nociceptors could also elicit an earlier negative component, labelled ultra-late N1. This ultra-late N1 would be similar to the N1 component preceding the Aδ-fibre N2-P2 complex. Indeed, such as the Aδ-related N1, the C-fibre related N1 was described as displaying a lateralized scalp topography, maximal over the temporal region contralateral to the stimulated side.
A small number of studies (Opsommer et al. 2001; Cruccu et al. 2003, Iannetti et al. 2003) have applied source analysis methods to C-fibre laser-evoked potentials. Such as Aδ-fibre laser-evoked potentials, C-fibre laser-evoked potentials were best explained by sources originating from bilateral opercular structures, as well as the anterior cingulate cortex. Hence, it would appear that Aδ and C fibre afferent volleys elicit brain responses within the same structures (see $ for a discussion on this topic).
3. Laser-evoked magnetic fields
The brain responses elicited by thermal stimulation of the skin have also been studied using MEG ($). Comparing laser-evoked brain responses recorded using EEG to those recorded using MEG could be interesting as the information recorded by both techniques may be complementary ($). For example, it is often stated that EEG signals mainly reflects sources of electro-cortical activity that are radial relative to the scalp surface, MEG signals mainly reflects sources of electro-cortical activity that are tangential relative to the scalp surface. In addition, deep sources of electro-cortical activity are thought to contribute very little to MEG signals, while superficial sources could be more accurately and more reliably captured using MEG. Indeed, the skull has a low conductivity to electric currents, while it is virtually transparent to magnetic fields. MEG recordings could thus be more spatially accurate than EEG recordings which are blurred and distorted by the interposed layers. However, it is important to highlight that these assertions have been questioned by a few recent studies (Malmivuo et al. 1997, 2004; Baumgartner 2004).
3.1 Laser-evoked magnetic fields elicited by the co-activation of Aδ and C nociceptors
Supporting the view that EEG and MEG provide information that is not redundant is the fact that the scalp topography and time course of laser-evoked magnetic fields is not identical to the scalp topography and time course of laser-evoked potentials. Indeed, laser-evoked magnetic fields mainly consist of an early response peaking at 150 ms when stimulating the hand dorsum and 200 ms when stimulating the foot dorsum (Kakigi et al. 1995), whose latency is similar to the laser-evoked N1 wave identified using EEG. In other words, it seems that MEG is very sensitive to the electro-cortical activity reflected in the N1 wave of laser-evoked potentials (and possibly other sources contributing at that latency), but is not very sensitive to the electro-cortical activity reflected in the later N2 and P2 waves of laser-evoked potentials.
A likely explanation as to why MEG does not capture the later cortical activity that is evident in laser-evoked potentials is that the recording technique is unable to capture activity originating from the cingulate cortex. Supporting this view, unlike source analyses of laser-evoked potentials, source analyses of laser-evoked magnetic fields have mostly failed to reveal responses originating from the cingulate cortex (Kakigi 1995, Watanabe 1998, Yamasaki 1999, Kanda 2000). This could be due to the predominantly radial orientation of this source of activity (Garcia Larrea 2003) and/or to its deep location. A notable exception is constituted by the findings of Ploner et al. ($) and Forss et al. ($), which did identify laser-evoked cingulate activity using MEG. However, the robustness of this claim is questioned by the fact that Forss et al. identified it in only 3/10 subjects, and the fact that the location and latency of the cingulate response identified by Ploner et al. was very unexpected, when considering what is usually identified using EEG.
While MEG studies have failed to consistently identify sources of activity originating from the ACC, they have shown more convincingly that laser stimuli can elicit electro-cortical activity originating from the contralateral primary somatosensory cortex (Kanda et al. 2000, Timmermann et al. 2001, Ploner et al. 2002, Raij et al. 2003, Forss et al. 2005). Interestingly, in most of these studies, the latency of the activity hypothesized to originate from the contralateral primary somatosensory cortex was concomitant or even slightly delayed as compared to the latency of the activity hypothesized to originate from the secondary somatosensory cortex.
3.2 Laser-evoked magnetic fields elicited by the selective activation of C fibres
A small number of studies have examined the MEG response elicited by the selective activation of C nociceptors (Tran et al. 2002, Kakigi 2003, Qiu 2004, Forss 2005). The elicited responses were hypothesized to reflect the bilateral activation of opercular regions. Furthermore, with the exception of the study by Forss et al., all these studies included a dipole whose location was compatible with activity originating from the contralateral primary somatosensory cortex. Therefore, the sources contributing to these C-fibre responses appear to be very similar to the sources contributing to the Aδ fibre responses.
4. Laser-evoked local field potentials (intra-cerebral recordings)
Several studies have examined whether noxious thermal stimuli applied to the skin elicit responses within different areas of the brain of awake humans, using surgically-implanted intracranial electrodes or subdural electrode grids ($). In all of these studies, laser stimulators were used to generate very brief (e.g. $ ms) stimuli above the thermal activation threshold of both C and Aδ nociceptors.
Intracranial recordings of laser-evoked brain responses have demonstrated that brief thermal stimuli elicit responses that can be recorded in the suprasylvian region. Using a subdural grid, Lenz et al. ($) recorded responses to laser stimuli and described a negative-positive wave that was maximal over the inferior aspect of the central sulcus, and that appeared to be generated in the vicinity of the suprasylvian opercular region, possibly, the insula or the deep vertical surface of the parietal operculum (i.e. not S2 proper). The latency of the response coincided with the latency of the N1 and N2 waves of laser-evoked potentials recorded using scalp EEG. Frot et al. ($) also identified activity within the suprasylvian opercular region, more specifically, in the parieto-frontal operculum. Latency of their responses was similar to that of the early N1 wave of laser-evoked potentials recorded using scalp EEG. Interestingly, they observed a delay of approximately 15 ms between the responses elicited in the ipsilateral and contralateral hemisphere. Furthermore, they showed that the thermal stimuli propably elicit two temporally-distinct responses within the suprasylvian region: an early response originating from the parietal operculum, followed by a later response originating from the insula. Using subdural grids, Vogel et al. ($) modelled the distribution of the elicited potentials and proposed a solution compatible with activity originating from the parieto-frontal operculum. Furthermore, Vogel et al. suggested that the somatotopic organization of the responses elicited by thermal stimulation was different from the somatotopic organization of the responses elicited by innocuous tactile stimulation, and, thereby concluded that both types of somatosensory stimuli were processed by distinct populations of operculo-insular neurons. In contrast, Frot et al. ($) did not find any significant difference between the spatial distribution of suprasylvian responses to nociceptive (laser) and non-nociceptive (electrical) stimulation of the upper limb ($).
Attempts have also been made to identify responses to thermal stimulation in the primary somatosensory cortex contralateral to the stimulated side. Using subdural grids, Kanda et al. ($) found that laser stimuli do elicit activity in the contralateral primary somatosensory cortex. However, contrary to the responses elicited by the electrical nerve stimulation of large-diameter Aβ fibres, the polarity of the identified response did not invert its phase across the central sulcus. This observation suggests that the elicited response was not generated in area 3b (tangential to the sulcus), but instead, in cortex largely radial to the subdural surface, for example, in areas 3a or 1.
In summary, intracerebral recordings have shown that nociceptive thermal laser stimuli elicit activity in bilateral opercular-insular regions, and, most probably, in the contralateral primary somatosensory cortex. Several studies have attempted to record laser-evoked potentials in other regions of the awake human brain. In a very recent study, Liu et al. (2009) showed that laser stimuli elicit consistent bilateral responses in the amygdala. Valeriani was unable to identify any clear response from deep intracerebral electrodes located around the thalamus ($). Using subdural grids, Lenz et al. ($) obtained reproducible biphasic responses in the mid portion of the anterior cingulate in three epileptic patients when stimulating the face, at a location compatible with BA24. However, for an unexplained reason, they failed to elicit similar responses when stimulating the upper limb. $ did succeed at recording reproducible responses to upper-limb stimulation using electrodes implanted in the mid anterior cingulate, i.e. BA24.
The latency of all these responses are always only compatible with the conduction velocity of myelinated Aδ fibres, as they are too early to be possibly conveyed by unmyelinated C fibres. However, most of these studies have restricted the time window of their analysis in such a way that it did not include latencies compatible with the conduction velocity of C fibres. Therefore, whether or not the concomitant activation of C fibres by the laser stimulus also elicits activity which can be captured using these intracranial electrodes or subdural electrode grids remains largely unknown.
Frot et al. (2007) examined the operculoinsular responses using different energies. Increasing stimulus intensities enhanced the magnitude of both the SII and the insular responses, and they claimed that the dynamics of the respective amplitude changes were different: S2 responses encoded gradually the intensity of stimuli, and tended to show a ceiling effect for higher intensities. In contrast, the insula did not respond reliably to low intensity laser pulses, but clearly encoded high stimulus intensities, without showing a similar ceiling effect.
It is often concluded that because these structures elicit measurable intra-cerebral laser evoked potentials, it is reasonable to assume that these structures contribute to the genesis of scalp laser-evoked potentials and laser-evoked magnetic fields. However, it is important to consider that this is not necessarily the case.
Some important points should be considered when interpreting the functional significance of these intracranial brain responses. (1) No studies have examined whether the concomitant activation or the selective activation of C nociceptors elicits any activity within these brain regions, and this algough it is clear that the laser stimuli elicit a percept of second pain. (2) No studies have examined whether other types of stimuli, e.g. a menacing visual stimulus, a loud auditory stimulus also elicit activity within these brain structures, within-subject, to determine the specificity of the elicited responses. (3) It is not known whether the magnitude of these responses is related to the intensity of the stimulus, or to the saliency of the stimulus, i.e. its ability to capture attention. (4) It is not known whether various experimental factors known to modulate the magnitude of laser-evoked potentials recorded from the scalp also modulate these responses in a similar fashion, which would constitute an important argument in favour of the claim that these responses contribute significantly to the scalp LEPs.
5. Important things to consider when interpreting the electrophysiological brain responses to the thermal activation of skin nociceptors
These different methods to sample the electrical brain activity elicited by the selective phasic activation of skin nociceptors offers a very exciting opportunity for researchers: it provides a method to study where, when and how nociceptive input is processed in the human brain, and how this processing may lead to the emergence of the pain experience. However, several important points should be considered when interpreting the functional significance of these responses.
(1) The elicited brain responses probably reflect only a fraction of the cortical activity generated by the nociceptive input.
(2) Because of the variability of the time required for heat conduction and transduction into a neural impulse ($), the nociceptive afferent volley generated by laser stimulation is probably not as synchronous as, for example, the non-nociceptive afferent volley triggered by the direct electrical activation of Aβ fibres (transcutaneous electrical activation). Furthermore, the target of the laser stimulus must be displaced from trial-to-trial and only a relatively small number of stimuli can be applied within a single experimental session to avoid skin overheating, nociceptor habituation and/or nociceptor sensitization. Therefore, laser-evoked brain responses are usually obtained in conditions that are far from optimal to identify signals that are very transient and/or of very small amplitude. This means that laser-evoked brain potentials probably reflect only a fraction of the electro-cortical activity generated by the nociceptive input.
(3) The magnitude of the elicited brain responses is often correlated with the energy of the eliciting stimulus, or with the intensity of the elicited percept. For this reason, investigators often conclude that these responses reflect brain processes related directly to the encoding of pain intensity. However, several studies have shown that this relationship may be disrupted, and, in fact, may be more related to the saliency of the stimulus, i.e. its ability to capture attention.
(4) The specificity of the eliciting stimulus does not preclude the specificity of the elicited brain responses.