The vertical dotted lines correspond to the latencies at which scalp maps ..... operculo-insular cortex of one patient (earlobe reference recording). The operculo-.
Université catholique de Louvain
Faculté de médecine Département de Physiologie et de Pharmacologie Unité de Neurophysiologie
Preceding events condition the central processing of nociceptive input as revealed by laser-evoked potentials
by André Mouraux, M.D.
Jury : Prof. Marc CROMMELINCK, President Prof. Léon Plaghki, Promoter Prof. Jean-Michel GUERIT, Co-promoter Prof. Giorgio CRUCCU Prof. Luis GARCIA-LARREA Prof. Bruno ROSSION Thèse présentée en vue de l’obtention du grade de Docteur en Sciences biomédicales Orientation: Neurosciences 2005
Cover figure: 3D representation of the time-frequency estimation of average EEG oscillation amplitude changes following CO2 laser stimulation. Laser stimulus concomitantly activated A - and C-fiber cutaneous nociceptors. Time-axis runs from left to right. Frequency-axis runs from back to front. See figure 3-3 for details.
2
Foreword A l’occasion de la présentation de ce travail, je voudrais tout d’abord remercier le Professeur Marc Crommelinck, président de mon jury, pour m’avoir accompagné tout au long de la préparation de cette thèse. Le Professeur Léon Plaghki, promoteur de la thèse, est l’initiateur de ce projet. Il a guidé mes premiers pas dans la voie de la recherche expérimentale, et n’a cessé de m’inspirer tout au long de ce travail. Il m’a accueilli comme étudiant-chercheur au laboratoire d’algologie en 1995. Je le remercie pour l’enseignement qu’il m’a prodigué et l’enthousiasme qu’il m’a communiqué. Le Professeur Jean-Michel Guérit, co-promoteur de la thèse, m’a soutenu dans ce projet dès le début de ma spécialisation en neurologie. Il s’est dévoué pour rendre possible ce travail de recherche. Je le remercie pour son soutien déterminant et amical. Le talent d’expérimentateur, la volonté, et l’enthousiasme du Professeur Bruno Rossion m’ont beaucoup aidé. Qu’il reçoive ici le témoignage de ma profonde amitié. Je dois un hommage particulier aux membres étrangers du jury, le Professeur Luis Garcia-Larrea et le Professeur Giorgio Cruccu, pour avoir influencé mes recherches. Leurs contributions scientifiques au domaine de la physiologie de la douleur sont aussi nombreuses qu’originales. Pour leurs commentaires constructifs, je remercie le Dr. Valery Legrain, mais aussi Mlle Céline Decruynaere et le Professeur Daniel Le Bars. Ceux-ci ont significativement remodelé l’interprétation de mes résultats expérimentaux. Pour la conception et la création du stimulateur thermique laser, outil indispensable à la réalisation des expériences de cette thèse, je remercie le Professeur André Fayt et Mr. l’abbé Stouffs. Le Dr. Laurent Bairy, le Dr. Quentin Verwacht, et Mr. François-Xavier Denijs ont également contribué à l’élaboration initiale de ce projet de thèse. Le Dr. Valérie Goffaux et Mr Corentin Jacques furent également des acteurs de ce projet. J’ai beaucoup profité de leur interaction dans la mise au point des outils de traitement du signal développés au cours de la préparation de ma thèse. Je remercie également Mlle. Adélaïde de Heering, Dr. Jean-François Delvenne, Mlle. Caroline Michel, Dr. Chrisine Schiltz, et Dr. Anne-Marie Schuller pour toutes ces discussions qui m’ont permis de progresser dans mon travail. Pour beaucoup, ces interactions professionnelles se sont muées en amitié. Je tiens à souligner la chaleur de l’accueil rencontré lors de mon arrivée dans l’unité NEFY. Je remercie tout particulièrement le Professeur Marcus Missal et le Professeur Etienne Olivier. Je remercie également Mr. Michaël Andres, Mr. Marco Davare, et Mlle Julie Duqué pour avoir si 3
souvent accepté de se prêter à mes expériences nociceptives mais aussi pour m’avoir fait découvrir les bienfaits de la stimulation magnétique transcrânienne. Je voudrais également exprimer ma gratitude envers Mlle. Coralie de Hemptinne et Dr. Alexandre Zénon pour m’avoir aidé à encadrer les travaux pratiques de neurophysiologie ainsi qu’envers Mme. Lilianne Lemmens et Mlle. Leila Azzaz pour leur gentillesse et leur disponibilité. Je dois également remercier le Professeur Christian Sindic ainsi que le Dr. Anne Jeanjean, le Dr. André Peeters, et le Dr. Michel Gille pour leur enseignement de qualité ainsi que pour m’avoir offert la possibilité de garder un contact avec la neurologie clinique pendant ces quatre années de recherche. Je remercie mes parents pour leur aide et leurs encouragements ainsi que mon grand-père, le Dr. Joseph Mouraux, pour m’avoir donné envie d’entamer des études de médecine. Enfin, pour tout l’amour qu’elles me donnent, mes remerciements s’adressent avant tout aux deux femmes de ma vie, Bénédicte et Manon.
4
Table of Contents
List of abbreviations .......................................................................................10 Chapter 1. Introduction ....................................................................................................13 1
General introduction.............................................................................................13
2
Nociceptive stimulators ........................................................................................16 2.1 Electrical nociceptive stimuli............................................................................16 2.2 Mechanical nociceptive stimuli ........................................................................17 2.3 Thermal nociceptive stimuli .............................................................................17 2.3.1
Laser heat nociceptive stimuli ...............................................................18
3
Sensations mediated by A - and C-fibers............................................................19
4
Laser-evoked brain potentials ..............................................................................25 4.1 The A -fiber mediated late LEP ......................................................................25 4.1.1
Latency, morphology, and topography ..................................................25
4.1.2
Hypothesized cortical generators ..........................................................31
4.1.2.1
Bilateral operculo-insular cortices ................................................33
4.1.2.2
Anterior cingulate cortex...............................................................37
4.1.2.3
Primary somatosensory cortex.....................................................39
4.1.2.4
Additional sources ........................................................................40
4.1.3
Comparing LEPs to laser-evoked responses obtained using other electrophysiological recording techniques.............................................41
4.1.3.1
Magnetoencephalography: LEPs vs. LEFs ..................................41
4.1.3.2
Intracranial recordings: LEPs vs. LFPs ........................................45
4.2 The C-fiber mediated ultra-late LEP................................................................47
5
4.2.1
Latency, morphology, and topography ..................................................47
4.2.2
Hypothesized cortical generators ..........................................................51
The ‘common generators’ hypothesis ..................................................................53 5.1 Late and ultra-late LEP responses may share common generators ...............53 5.2 Vertex potentials..............................................................................................53 5.2.1
Vertex potentials in the auditory modality..............................................55
5.2.2
Vertex potentials in the somatosensory modality ..................................56
5.2.3
Common processes underlying vertex potentials..................................58
5.3 Nociceptive specificity of laser-evoked potentials ...........................................58 5.3.1
Nociceptive specificity of the laser-evoked N2 potential........................58
5
6
5.3.2
Nociceptive specificity of the laser-evoked P2 potential........................60
5.3.3
Nociceptive specificity of the laser-evoked N1 potential........................61
Experimental modulation of late LEPs, ultra-late LEPs, and vertex potentials.....64 6.1 Stimulus intensity – Intensity of perception .....................................................64 6.1.1
Late LEPs ..............................................................................................64
6.1.2
Ultra-late LEPs ......................................................................................64
6.1.3
Somatosensory and auditory vertex potentials......................................66
6.2 Stimulus repetition – Interstimulus interval......................................................66 6.2.1
LEPs ......................................................................................................66
6.2.2
Somatosensory and auditory vertex potentials......................................67
6.3 Cognitive and attentional factors .....................................................................71 6.3.1
Vigilance ................................................................................................71
6.3.1.1
Vertex potentials...........................................................................71
6.3.1.2
LEPs.............................................................................................73
6.3.2
Selective attention .................................................................................75
6.3.2.1
Selective attention within the auditory modality............................76
6.3.2.2
Selective attention within the somatosensory modality ................78
6.3.2.3
Inter-modal selective attention .....................................................78
6.3.2.4
Selective attention within the nociceptive modality ......................79
6.3.3
Task relevance ......................................................................................87
6.3.3.1 6.3.4
The laser evoked P600 ................................................................87
Stimulus-driven attentional capture .......................................................89 6.3.4.1.1 The laser-evoked “P400 effect” ..............................................91
6.4 In brief .............................................................................................................93 7
Why C-nociceptive input elicits an ultra-late LEP only when concomitant activation of A -fibers is avoided..........................................................................94 7.1 Non-stationarity of the C-fiber afferent volley ..................................................94 7.2 A -fiber mediated spinal inhibition of C-fiber afferent transmission ................95 7.3 Refractoriness of LEP cortical generators.......................................................96
Chapter 2. Study Objectives ............................................................................................97 Chapter 3. Non-phase locked EEG responses to CO2 laser skin stimulations may reflect central interactions between A - and C-fiber afferent volleys ........................99 Abstract ....................................................................................................................100 1
Introduction ........................................................................................................101
2
Methods .............................................................................................................103
6
2.1 Subjects.........................................................................................................103 2.2 Test stimulus and CO2 laser stimulator .........................................................103 2.3 Experimental design......................................................................................104 2.4 Data acquisition.............................................................................................104 2.4.1
Reaction time and intensity of perception............................................104
2.4.2
Electroencephalogram.........................................................................104
2.5 Data analysis.................................................................................................105 2.5.1 3
Time-frequency transformation of the data..........................................105
Results ...............................................................................................................106 3.1 Intensity of perception and reaction time.......................................................106 3.2 Evoked potentials ..........................................................................................107 3.3 Time-frequency analysis ...............................................................................108 3.3.1
3.3.1.1
ROI 1 or ‘Late LEP’ ....................................................................111
3.3.1.2
ROI 2 or ‘Ultra-late LEP’.............................................................112
3.3.2
Non phase-locked induced activities ...................................................112
3.3.2.1
ROI 3 or ‘Late ERS’....................................................................114
3.3.2.2
ROI 4 and 5 or ‘Late and Ultra-late ERD’ ...................................114
3.3.3 4
Phase-locked evoked changes............................................................111
Overview of results ..............................................................................115
Discussion..........................................................................................................118 4.1 Using reaction-time to distinguish between A - and C-fiber mediated detection 118 4.2 Stimulus related EEG changes revealed by time-frequency wavelet transform 119
5
4.2.1
A - and C-fiber related late and ultra-late LEPs..................................119
4.2.2
A - fiber related ‘Late ERS’ .................................................................120
4.2.3
A - and C fiber related late and ultra-late ERD ...................................123
Conclusion .........................................................................................................124
Chapter 4. Refractoriness cannot explain why C-fiber laser-evoked brain potentials are recorded only if concomitant A -fiber activation is avoided ......................... 127 Abstract ....................................................................................................................127 1
Introduction ........................................................................................................128
2
Methods .............................................................................................................129 2.1 Subjects.........................................................................................................129 2.2 Stimulus.........................................................................................................129 2.3 Experimental design......................................................................................130
7
2.4 Data acquisition.............................................................................................131 2.4.1
Intensity of perception .........................................................................131
2.4.2
Electroencephalogram.........................................................................131
2.5 Data analysis.................................................................................................132
3
2.5.1
Artifact correction and rejection ...........................................................132
2.5.2
Average and subtraction waveforms ...................................................132
2.5.3
Peak extraction....................................................................................133
2.5.4
Statistical analysis ...............................................................................133
Results ...............................................................................................................134 3.1 Probability of detection ..................................................................................134 3.2 Intensity of perception ...................................................................................135 3.3 Event-related potentials.................................................................................136
4
3.3.1
N2 and P2 components .......................................................................136
3.3.2
P600 component .................................................................................140
Discussion..........................................................................................................144 4.1 Testing the ‘refractory period’ of LEPs ..........................................................144 4.2 P600 component ...........................................................................................146 4.3 Where is the ultra-late LEP? .........................................................................148
5
Conclusion .........................................................................................................149
Chapter 5. Are laser-evoked brain potentials modulated by attending to first or second pain? ............................................................................................................ 151 Abstract ....................................................................................................................151 1
Introduction ........................................................................................................152
2
Methods .............................................................................................................153 2.1 Subjects.........................................................................................................153 2.2 Test stimulus .................................................................................................153 2.3 Preliminary procedures .................................................................................154 2.4 Training session ............................................................................................155 2.5 Recording session.........................................................................................155 2.6 EEG acquisition and analysis........................................................................156
3
Results ...............................................................................................................157 3.1 Behavioral results..........................................................................................157 3.2 Electrophysiological results ...........................................................................159
8
3.2.1
Late N2-P2 complex ............................................................................159
3.2.2
Late P600 component .........................................................................159
3.2.3
Ultra-late N2-P2 complex ....................................................................160
4
Discussion..........................................................................................................161 4.1 Detection latency and sensation qualities of first and second pain ...............161 4.2 Modulation of LEPs by attending to first or second pain. ..............................161
5
Conclusion .........................................................................................................162
Chapter 6. Further considerations ................................................................................. 163 1
Saliency of A - and C-fiber input........................................................................165
2
Expectancy.........................................................................................................166
Chapter 7. Conclusion and future perspectives ............................................................. 171 Appendix A. Time-frequency analysis of event-related potentials using the wavelet decomposition of EEG epochs..................................................................... 175 Time-domain averaging and the ‘additive noise’ model ...........................................175 ERD and ERS ..........................................................................................................177 Revealing ERD and ERS .........................................................................................178 Continuous Morlet wavelet transform of EEG epochs .............................................180
Appendix B. Single-trial detection of human brain responses evoked by laser activation of A -nociceptors using the wavelet transform of EEG epochs ....................... 183 Appendix C. LetsWave: an EEG signal-processing toolbox written in Borland Delphi ..... 193 Appendix D. Publications related to this thesis................................................................. 197 References.................................................................................................. 199
9
List of abbreviations
ACC
Anterior cingulate cortex
AMH
A -mechano-heat receptor
ANOVA Analysis of Variance
10
BA
Brodmann area
CMH
C-mechano-heat receptor
CMT
Continuous Morlet transform
CNV
Contingent negative variation
CWT
Continuous wavelet transform
EEG
Electroencephalogram
EOG
Electro-oculogram
ERD
Event-related desynchronization
ERP
Event-related potential
ERS
Event-related synchronization
ERV
EEG response value
FFT
Fast-Fourier transform
fMRI
Functional magnetic resonance imaging
IC
Independent component
ICA
Independent component analysis
ISI
Inter-stimulus interval
LEF
Laser-evoked magnetic fields
LEP
Laser-evoked brain potential
LFP
Local field potentials
MEG
Magnetoencephalography
MMN
Mismatch negativity
MRCP
Motor-related cortical potentials
MRI
Magnetic resonance imaging
Nd
Negative difference
PET
Positron emission tomography
PN
Processing negativity
ROC
Receiver-operating characteristic
ROI
Region of interest
RT
Reaction-time
SEP
Somatosensory-evoked potentials
SI
Primary somatosensory area
SII
Secondary somatosensory area
SNR
Signal-to-noise ratio
SOA
Stimulus onset asynchrony
TF
Time-frequency
VAS
Visual-analogue scale
11
12
Chapter 1. Introduction
1 General introduction Pain, defined as a percept, is a complex and primarily subjective experience which
involves
multidimensional
sensory,
motivational,
and
cognitive
components. The sensory system producing this perception, sometimes referred to as the ‘nociceptive’ system, consists in cutaneous and visceral nociceptors (present in all tissues except the brain), peripheral A - and C-fiber afferent fibers, and spinal transmission neurons which modulate and project this peripheral input to supraspinal structures such as the brain stem, the thalamus, the limbic system, and the cortex. A vital function of the nociceptive system is to provide immediate awareness of threats to the body’s integrity, inciting the individual to react by producing an adequate protective response. Therefore, for nociceptive events to interrupt ongoing behavioral goals, nociceptive brain processes are expected to be strongly interlaced with attentional processes. A pertinent approach to the study of the nociceptive system in humans is therefore to employ non-invasive observational and experimental methods which give access to nociceptive-related brain processes at an integrative level of the central nervous system. Event-related brain potentials (ERPs) have provided substantial information regarding sensory and cognitive processes which may occur in the brain. ERPs appear as transient changes in the ongoing electrical brain activity, time-locked to the onset of the event. ERPs are hypothesized to result from sudden and synchronized increases of postsynaptic activity, occurring in similarly oriented neuronal populations. To extract evoked potentials from the ongoing, non event-related, electrical brain activity, the event is usually repeated such as to allow the averaging of successive peristimulus EEG recordings. The principle
underlying these techniques is that averaging
successive EEG epochs should cancel out the contribution of signals which are not ‘time-locked’ or ‘stationary’ to the onset of the event while it should preserve evoked activity which is assumed to occur with a constant timedelay. The fraction of the signal which is cancelled-out by the averaging 13
procedure is often referred to as ‘additive noise’ (Regan 1989; see Appendix A). Event-related potentials typically consist of a series of voltage polarity changes, observed as peaks and troughs in the average waveform. These potentials can be classified according to their relative timing to stimulus onset, their polarity, and their magnitude. In most cases, each individualized ERP deflection corresponds to neural activity arising from several temporally overlapping sources. As ERPs provide a high temporal resolution, they can be used to distinguish and identify the different neural processes involved in perceptual tasks. Indeed, depending on their modality, sensory stimuli elicit a series of sensory or exogenous ERP peaks which reflect the initial processing occurring in modality-specific cortical areas. Following these peaks, later components may be recorded, which are thought to reflect more integrative and endogenous aspects of perception. The understanding of the cortical processes underlying pain perception is well behind that of other sensory modalities. The difficulties related to the production of an adequate and selective nociceptive stimulus most probably contributed to this setback. In 1975, Mor and Carmon introduced the infrared laser stimulator. Allowing brief, synchronous, and selective activation of cutaneous A - and C-fiber nociceptors, laser heat stimulators have, since then, been extensively used to study time-locked nociception-related behavioral and electrophysiological responses. The characteristics of infrared laser stimulators will be discussed in section 2 of Introduction. Laser stimuli which concomitantly activate A - and C-fiber nociceptors produce a characteristic double sensation, reminiscent of the ‘first’ and ‘second’ pain described by Lewis and Ponchin (1937). First pain is often described as a localized and short-lasting ‘pricking’ sensation while second pain is often described as a ‘burning’ sensation which spreads beyond the spatial and temporal limits of the stimulus. The perceptual correlates of A and C-nociceptor activation will be discussed in section 3 of Introduction. Although subjects clearly report sensations related to the activation of both A - and C-fiber nociceptors, laser-evoked brain potentials (LEPs) have only 14
revealed components whose latencies are compatible with the conduction velocity of A -fibers (i.e. the ‘late LEP’; ~160 – 390 ms; Bromm and Treede, 1984). Several methods allow narrowing the selectivity of the laser stimulator such as to activate C-fibers in isolation (Plaghki and Mouraux 2002). Most curiously, all these methods have shown that avoiding the concomitant activation of A -fibers not only resulted in the disappearance of first pain and its electrophysiological correlate, the late LEP, but also led to the appearance of an ultra-late LEP whose latency (~750 – 1150 ms) was compatible with the arrival time of C-fiber input. As discussed in section 4 of Introduction, late and ultra-late
LEP
waveforms
have
markedly
similar
morphologies
and
topographies. What’s more, both waveforms may be explained by similar source configurations. Furthermore, as reviewed in section 6 of Introduction, late and ultra-late LEP responses appear to be equally modulated by experimental manipulations of the subject’s experimental surroundings, general level of arousal, and focus of attention. For these reasons, several investigators have proposed that both responses could reflect the activation of a cortical network common to the processing of both afferents. Furthermore, a number of investigators have pointed at the resemblance between LEP components and late ‘vertex potentials’ which may be elicited by stimuli of all sensory modalities. This and its implications regarding the functional significance and nociceptive specificity of the cortical processes underlying LEPs will be discussed in section 5 of Introduction. Why the C-fiber afferent volley does not elicit an ultra-late LEP response when it shortly follows an A -fiber afferent volley is a poorly understood phenomenon. Furthermore, a similar phenomenon might be related to why investigators have failed to record reproducible A -fiber mediated evoked potentials when A -fibers are concomitantly activated (Boulu et al. 1985; De Broucker and Willer 1985). In other words, one could hypothesize that similar mechanisms condition the occurrence of both late and ultra-late LEPs. A review and critique of the different explanations which have been put forward to account for these observations constitutes section 7 of Introduction.
15
Whatsoever, the finding that activating A - or C-fiber nociceptors can produce sensations without necessarily evoking late or ultra-late LEP responses should be taken into account when one infers on the functional significance of the processes these responses reflect. Indeed, the apparent dissociation between perceptual and electrophysiological correlates of A - or C-nociceptor activation would suggest that LEPs reflect cortical processes which are not required for the perception of first or second pain. Therefore, it is most likely that late and ultra-late LEP responses only reflect a fraction of the cortical processing of both nociceptive inputs. 2 Nociceptive stimulators Both in fundamental and clinical research, the study of sensory systems requires providing input to that system in the form of a perfectly controlled stimulus. It is probably because instruments producing such stimuli have been readily available to the scientific and medical community that current knowledge about visual and auditory sensory systems has reached such a remarkable level today. Any experimental nociceptive stimulus should be quantifiable (i.e. defined in physical units as to intensity, time, and spatial distribution), reproducible, and safe. Furthermore, in order to selectively activate the nociceptive system, the stimulus should predominantly activate A - and C-fiber nociceptors (Gasser and Erlanger 1929). Indeed, concurrent activation of other sensory modalities, such as A -fiber afferents of the lemniscal pathway, should be avoided as their activation could produce overlapping responses and, more importantly, modulate the nociceptive responses themselves. Electrical, mechanical, and thermal stimuli all fulfill some of these requirements. However, all have their specific shortcomings. 2.1
Electrical nociceptive stimuli
The advantage of electrical stimulation is that the stimulus is easily controlled and implemented. For these reasons, electrical stimuli have often been used to produce nociception in both animal and human studies. However, if an electrical stimulus is used to activate A - and C-fiber afferents, concurrent activation of other, non-nociceptive, afferent fibers is unavoidable. 16
Indeed, due to their large diameter, fast-conducting myelinated afferent fibers which mostly mediate input arising from mechano-receptors have lower electrical impedance, and therefore lower electrical activation thresholds, than the thinly myelinated A -fibers and the unmyelinated C-fibers which convey nociceptive input. Intracutaneous stimulation electrodes (Bromm and Meier 1984; Inui et al. 2002) and intraneural microstimulation electrodes (Torebjork and Ochoa 1980) have been developed in an attempt to circumvent this problem. However, these methods each have their own technical limitations (Handwerker and Kobal 1993; Treede 1994). Finally, it should be emphasized that electrical activation of afferent fibers bypasses the processes related to receptor transduction. While this prevents studying these processes, it may also allow a better synchronization of the afferent input. 2.2
Mechanical nociceptive stimuli
Mechanical stimuli delivered with needles or pressure algometers are commonly used tools for clinical evaluation. However, such as electrical stimulation, these stimulation methods lack selectivity as they concurrently activate mechanoreceptors and nociceptors. Furthermore, most conventional mechanical stimulators do not provide the speed and precision required for the study of time-locked psychophysical and electrophysiological events. The use of high-energy ultrasound is also problematic as the biophysical effects are not well understood (Gavrilov et al. 1977). Indeed, it is not clear whether the sensory effect of such stimuli is produced by thermal or mechanical energy. Finally, it should be noted that controlled impact of small metallic cylinders may provide fast stimulus onsets (Kohlloffel et al. 1991). However, these latter methods have not been extensively used in psychophysical studies. 2.3
Thermal nociceptive stimuli
Heat is the most frequently used form of natural noxious stimulation. However, if the source of energy is heat conducted from a thermode or radiant heat such as that produced by a light bulb, the usefulness of this stimulation 17
method is limited. Indeed, the rise in cutaneous temperature produced by these stimulators is too slow to produce afferent inputs sufficiently synchronous to allow the study of time-locked events such as reaction-times, muscle reflexes, neuronal responses, and event-related brain potentials. Another disadvantage of conventional radiant heat sources is that their energy is emitted in visible and near infrared regions of the electromagnetic spectrum. At these wavelengths, the energy absorption of the skin is poor. Furthermore, reflectivity is important, and among other factors, depends on skin pigmentation. Consequently, the overall control of heat transfer is fairly unpredictable. Thermodes have another disadvantage related to the fact that they require a contact with the skin. Indeed, applying the thermode against the skin concomitantly activates low threshold mechanosensitive afferent fibers, thereby reducing the nociceptive selectivity of such stimulators. Furthermore, the rigid and planar surfaces of thermodes limit their usability as most cutaneous surfaces are not flat. 2.3.1 Laser heat nociceptive stimuli Monochromatic radiant heat sources of high power density such as infrared lasers are able to circumvent most of the shortcomings encountered by conventional heat stimulators. In comparison to classical incandescent light sources which emit their radiative energy in all spatial directions and in a broad spectrum of wavelengths, the laser energy is confined to a narrow beam of nearly parallel monochromatic
electromagnetic
waves.
The
combination
of
these
characteristics allows lasers to produce stimuli whose spectral energy density is several orders of magnitude greater than that produced by conventional light sources. The cutaneous temperature ramps produced by a laser stimulator may rise to several thousands of degrees per second. The activation of cutaneous nociceptors is therefore sufficiently synchronized to allow the recording of time-locked neural responses. Furthermore, unlike thermodes, laser stimulators do not require a contact with the skin.
18
As pioneered by Mor and Carmon (1975) high-power CO2 lasers have characteristics which make them very appropriate cutaneous heat stimulators (for a discussion on other classes of infrared lasers, see Arendt-Nielsen and Chen 2003). The CO2 laser emits in the far infrared (10.6 µm). At such wavelengths, absorption of the skin is nearly complete and transparency is not only very low but also independent of skin pigmentation (Hardy and Muschenheim 1934; see figure 1-1). For these reasons, the calorific energy of the stimulus remains confined to the upper skin layers, where transducer nerve terminals sensitive to thermal variations are located. Furthermore, there is no need for invasive intracutaneous temperature measurements as temperature profiles can easily be modeled in both time and space.
visible
Fig. 1-1. Skin reflectance as a function of wavelength is shown for both white skin (dashed line) and pigmented skin (solid line). At wavelengths greater than 2 µm, skin reflectance is independent of pigmentation (adapted from Hardy and Muschenheim, 1934).
3 Sensations mediated by A - and C-fibers Well controlled, brief, and powerful laser stimuli directed to a non glabrous area of the skin provide an experimental mean of producing selective and quasi-synchronous activation of A - and C-fiber nociceptors. 19
Albeit it’s nociceptive specificity, brief laser stimuli applied to the hairy skin (e.g. dorsum of the hand) do not necessarily evoke a painful sensation (Bromm and Meier 1984; Svensson et al. 1997; Nahra and Plaghki 2003b). Indeed, at stimulus intensities slightly above detection threshold, perception is dominated by warmth and touch-like sensations which are detected with latencies above 800 ms. At higher intensities, perception of the stimulus is described as a well-localized, predominantly painful, pricking, and burning sensation. This sensation is detected with much shorter reaction-times (~350 ms). Most often, it is followed by a second, diffuse, and long-lasting, warm or burning sensation. This double sensation is remnant of the ‘first pain’ and ‘second pain’ characterized by Lewis and Ponchin (1937). The dual nature of these perceptual responses may be interpreted as resulting from the activation of two distinct afferent nociceptive pathways. One with a low heat threshold (~40 °C) and low conduction velocity (~1 m/s) related to nonmyelinated C-fibers, the other with a high heat threshold (~46 °C) and fast conduction velocity (~10 m/s) related to small myelinated A -fibers (Bromm et al. 1984; Bjerring and Arendt-Nielsen 1988; Nahra and Plaghki 2003b). Indeed, these psychophysical and neurophysiological studies have shown that the two perceptual responses are respectively related to the activation of A and C-fiber primary afferents. To differentiate between stimulation of the glabrous and the hairy skin is of importance. Indeed, heat stimuli applied to the glabrous skin of the human hand do not appear to produce sensations similar to the first pain sensation produced by stimulation of the hairy skin. This observation may be related to the failure to find A -mechano-heat (AMH) type II receptors in the glabrous skin of the monkey (Treede et al. 1995). Indeed, AMH type II receptors are most probably the main mediators of the laser-evoked sensation of first pain as (1) the latency of the response to first pain is too short to be related to slowly-conducting C-fibers, (2) the threshold of AMH type II receptors is very close to that of first pain (Dubner et al. 1977), and (3) the burst of AMH type II receptor activity at stimulus onset is consistent with the perception of a shortlasting pricking sensation (Campbell and Lamotte 1983). In contrast, Meyer 20
and Campbell (1981) have shown that AMH type I receptors are mostly involved in the perception of long-lasting tonic heat stimuli (see figure 1-2).
Fig. 1-2. Ratings of pain by human subjects during a long-duration (30 s), intense heat (53 °C) stimulus applied to the glabrous hand are compared with responses of CMH and type I AMH. A. Pain was intense throughout the stimulus (n = 8). B. The brisk response of the CMH at the beginning of the stimulus changed to a low rate of discharge after 5 s (n = 15). C. The response of type I AMH increased during the first 5 s and remained high throughout the stimulus (n = 14). AMH: A-fiber mechano-heat receptors. CMH: C-fiber mechano-heat receptors. From Meyer and Campbell (1981).
21
Fig. 1-3. A. Co-activation of A - and C-nociceptors. The difference in conduction velocity of A - and C-fibers explains why subjects most often report a double sensation (i.e. first and second pain). Indeed, the A -fiber afferent volley arrives much faster at central projection sites than the slower C-fiber afferent volley. Reaction-times reflect the latency of detection of the A -fiber mediated sensation. B. Selective activation of C-nociceptors. In this example, an ischemic A-fiber block of the superficial radial nerve was used to selectively activate C-fiber afferents. Subjects reported the disappearance of first pain but the persistence of a delayed sensation of second pain. Reaction-times reflect the latency of detection of slowly-conducting C-fibers.
22
The difference between detection latency of A -fiber mediated first pain and C-fiber mediated second pain is only observed when very fast heating ramps are used. Indeed, only under these conditions will the stimulus activate all receptors quasi-simultaneously, and this despite the fact that the heat threshold of A -nociceptors is higher than that of C-nociceptors. Due to the difference between A - and C-fiber conduction velocities, A -nociceptive input will reach central projections well before C-nociceptive input (see panel A of figure 1-3). Depending on peripheral conduction distance, the difference in arrival-time of both afferent volleys may thus vary from 0.1s (e.g. stimulation of the trigeminal area) to more than 1 second (e.g. stimulation of the foot dorsum). This delay has been extensively exploited by investigators to discriminate between responses elicited by the activation of A -fibers from that elicited by the activation of C-fibers. It is commonly accepted that A - and C-fibers mediate different aspects of pain perception. A -fibers convey sensations which are usually described as well-localized, sharp, or pricking, and appear to be mostly implicated in phasic pain. The faster conduction velocity of A -fibers agrees with the often adopted view that A -fiber input serves primarily as an early warning signal. In contrast, slow unmyelinated C-fibers convey sensations which appear more implicated in tonic pain and are generally associated with adaptive selfprotective behaviors. Indeed, C-fiber mediated sensations are usually described as warm, burning, or aching sensations spreading well beyond the spatial and temporal limits of the stimulus. A number of psychophysical studies have compared the perception of nociceptive heat before and after applying an ischemic nerve conduction block selectively affecting myelinated A-fibers (Landau and Bishop 1953; Sinclair and Stokes 1964; Price et al. 1977). Most of these studies have shown that Cfibers were the main mediators of heat pain. These studies all shared in common the use of tonic heat stimuli of long duration and relatively large surface areas. As compared to these studies, the laser stimulator produces a phasic nociceptive stimulus of much shorter duration and much smaller surface area. Several studies have examined the relative contribution of A 23
and C-fibers to the perceptions evoked by such a laser stimulus (Chakour et al. 1996; Nahra and Plaghki 2003b). These studies have shown that the sensation produced by brief and localized laser heat stimuli was mainly mediated by A -fibers. Indeed, when A -fiber afferents were blocked (e.g. by applying a prolonged ischemic pressure block against the superficial radial nerve; see panel B of figure 1-3), discrimination performance, solely relying on the remaining C-fiber mediated input, was greatly reduced. Consequently, blocking A -fiber afferents resulted in an important increase in absolute detection threshold (see figure 1-4). The sparse contribution of C-fibers to the perception of laser stimuli could be due to C-fiber afferents requiring an important temporal and spatial summation to evoke consistent perception (Price et al. 1977). From these observations, it appears that long-lasting tonic heat stimuli produce sensations which are mostly mediated by C-fibers, while brief phasic laser heat stimuli produce sensations which are mostly mediated by A -fibers. Control
A-fiber block
A.
B.
C.
Stimulus strength (mJ/mm 2)
24
Fig. 1-4. Effect of A-fiber ischemic pressure block on detection performance, magnitude of perception and reaction-times to CO2 laser stimuli applied to the dorsum of the left hand (energy density: 5.8 - 10.6 mJ/mm2; duration: 40 ms; stimulus surface area: 20 mm2). A. Probability of detection as a function of stimulus strength. After A-fiber block, probability of detection was significantly reduced. B. The intensity of perception, measured on a visual-analogue scale, increased monotonically with stimulus strength in both conditions. However, after A-fiber block, the curve was much less steep. Consequently, 39.3% of all presented stimuli were perceived as painful in the control condition against 9.7% in the A-fiber block condition. C. In the control condition, reaction-times decreased with stimulus strength. During A-fiber block, almost all reactiontimes were of long latency (from Nahra and Plaghki, 2003).
4 Laser-evoked brain potentials The synchronous and concomitant activation of A - and C-fiber nociceptors, resulting from the very steep heating ramps produced by CO2 laser heat stimulators, allows the recording of event-related brain potentials. Laserevoked brain potentials (LEPs) have revealed components whose latencies are compatible with the conduction velocity of A -fibers (i.e. the ‘late LEP’: ~160 – 390 ms; e.g. Figure 1-5A). Although subjects clearly report the perception of both A -fiber related first pain and C-fiber related second pain, no evoked potentials are recorded at latencies compatible with the conduction time of C-fibers. At first, the absence of C-fiber related ultra-late components was assumed to result from the non-stationarity of the evoked afferent volley (discussed in section 7.1 of Introduction). However, avoiding the concomitant activation of A -fibers does not only lead to the disappearance of first pain and its electrophysiological correlate, the late LEP but also leads to the appearance of an ‘ultra-late LEP’ whose latency (~ 750 – 1150 ms) is compatible with the arrival time of C-fiber input (e.g. Figure 1-5B). 4.1
The A -fiber mediated late LEP
4.1.1 Latency, morphology, and topography The most prominent component of the LEP response mediated by A -fibers consists of a large, biphasic, negative-positive complex (N2-P2) culminating at the vertex (Figure 1-5A). Due to the relatively slow conduction velocity of A fibers (~10 m/s), the latency of this complex is significantly influenced by peripheral conduction distance. When stimulating the trigeminal area, the latency of late N2 and P2 components are respectively 160 – 180 and 240 – 270 ms (for a review of trigeminal responses to laser stimuli, see Romaniello et al. 2003). When stimulating the dorsum of the hand, the latency of these components increases to respectively 250 and 350 ms. When stimulating the dorsum of the foot, the latency of these components is further delayed, occurring respectively around 350 and 400 ms after stimulus onset (for a review, see Chen et al. 1998).
25
A.
N2 CZ CONTROL
10 µV + P2
B.
A-FIBER BLOCK
N2
CZ
500 ms
P2 0
LLEP
ULEP
Fig. 1-5. Laser-evoked potentials (LEPs) recorded in 9 subjects before (panel A) and after (panel B) applying an ischemic A-fiber pressure block to the superficial radial nerve (grand-average; A1A2 reference). Four different stimulus intensities, ranging from 5.8 to 10.6 mJ/mm2 were used (labeled ‘1’ to ‘4). LLEP: the time-window within which A -fiber related late LEP components are usually recorded after stimulation of the hand (160 - 390 ms). ULEP: the time-window within which C-fiber related ultralate LEP components are usually recorded (750 - 1150 ms). Note that unlike the amplitude of the late LEP recorded in the control condition, the amplitude of the ultralate LEP recorded in the A-fiber block condition was mostly uncorrelated with stimulus intensity (adapted from Nahra and Plaghki, 2003).
26
The P2 component (Treede et al. 1988a; Miyazaki et al. 1994; Xu et al. 1995; Valeriani et al. 1996) displays a central and widespread central scalp topography whose maximum is recorded at the vertex (electrode CZ; see figure 1-6). Such as the P2 component, the N2 component (Treede et al. 1988a; Kunde and Treede 1993) is also maximal at the vertex but extends bilaterally to lateral sites of the scalp (temporal electrodes T3 and T4; see figure 1-6). A contralateral predominance has sometimes been described (Kanda et al. 1999). In a study examining LEP responses elicited by stimulation of the dorsum of the foot, Valeriani et al. (1996) proposed that the laser-evoked N2 component could reflect the superposition of two distinct but temporally overlapping subcomponents, labeled N2a and N2b. As compared to the earlier N2a component, the N2b component would display a more frontal and medial topography (see Figure 1-7).
Fig. 1-6. Laser-evoked potentials (LEPs) were recorded in 12 subjects. Stimulus, applied to the dorsum of the left hand, was above the threshold of both A - and Cnociceptors (9.5 ± 0.5 mJ/mm2; 40 ms duration; 10 mm diameter; ISI 10 – 20 s). Solid waveform: grand-average obtained at electrode CZ vs. A1A2. Dashed waveform: grand-average obtained at contralateral electrode T4 vs. FP1.
27
Several studies (Treede et al. 1988a; Kunde and Treede 1993; Miyazaki et al. 1994; Xu et al. 1995; Spiegel et al. 1996) have shown that the laser stimulus can evoke an additional and earlier negativity, labeled N1 (see Figures 1-6 and 1-7). The N1 component precedes the late vertex N2 component and is often described as “riding on the ascending N2 negativity” (Treede et al. 1988a). When stimulating the dorsum of the hand, the latency of the N1 component is approximately 170 ms. The topographical distribution of the N1 component is different from that of the N2 component. Indeed, the N1 component is lateralized, being maximal at temporal leads contralateral to the stimulation site. N2 and P2 components are usually best identified using linked earlobes as reference. To identify the N1 component and dissociate it from the partially overlapping N2 component, a frontal median reference electrode is most often used (Kunde and Treede 1993; Valeriani et al. 1996; Valeriani et al. 2000b). Indeed, the positive counterpart of the N1 component, sometimes labeled P1, may be recorded at such scalp locations. The significant correlations between N1 and P1 amplitudes and latencies is indeed a strong indication that this P1 component is the positive counterpart of the electrical brain activity underlying the N1 and not a distinct laser-evoked component. It should be noted that Spiegel et al. (1996) described an additional ipsilateral N1 component, of lower amplitude. However, this result could have been related to the P1 positivity affecting the frontal-median reference electrode used in this study. Xu et al. (1995) described a shift in topography of the N1 component when changing stimulation site. Indeed, these investigators 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 topography was not found in other studies examining LEPs elicited by stimulation of different body locations (Tarkka and Treede 1993; Spiegel et al. 1996; Valeriani et al. 1996).
28
Fig. 1-7. Laser-evoked potentials (LEPs) to left hand stimulation in one subject. Average waveforms were recorded at the right eye referred to the nose (EOG), T4 referred to FZ, and T4, T3, FZ, F4, CZ referred to the nose. Two superimposed averages of 100 trials are displayed. The vertical dotted lines correspond to the latencies at which scalp maps were obtained by linear interpolation. N1, N2, and P2 components could clearly be identified in the average waveforms. Based on topographical differences, investigators proposed that the N2 component was composed of two distinct subcomponents labeled N2a and N2b. The N2a potential (map B) was maximal at the vertex and on left temporal regions. The N2b potential (map C) was mostly distributed at right frontal regions (from Valeriani et al., 1996).
Valeriani et al. (2000b) suggested that the laser stimulus could, in certain circumstances, elicit an even more precocious LEP component, occurring approximately 80 ms after stimulating the hand dorsum, and referred to as ‘eP’ or ‘early positivity’ (see Figure 1-8). This very early component displayed a contralateral temporal topography and appeared to be recorded only when subjects were required to perform a point-discrimination task. However, it
29
should be emphasized that several studies using similar spatial-discrimination tasks have failed to reproduce this finding (e.g. Kanda et al. 1999).
Fig. 1-8. Grand-average of laser-evoked potentials recorded in ten subjects. CO2 laser stimuli were randomly delivered at three different skin regions of the right hand dorsum. After each trial, subjects were asked to identify the stimulated region (spatial discrimination task). Previous to grand-averaging, all traces were made coincident at the peak latency of the N1 potential. An early positive response (eP) was identified at the T3 trace (nose reference). Scalp maps, calculated at eP, N1/P1, N2a, and P2 latencies, are shown on the right. The scalp topography of the eP component was different from that of the N1/P1 component (adapted from Valeriani et al., 2000).
30
4.1.2 Hypothesized cortical generators A number of studies have applied source analysis methods to the electrical scalp activity elicited by cutaneous laser stimuli. Most of these studies have used methods based on the optimization of a fixed spatio-temporal dipole configuration using a spherical head model (Tarkka and Treede 1993; Bromm and Chen 1995; Valeriani et al. 1996; Valeriani et al. 2000b; Schlereth et al. 2003). More recently, a few number of studies have used more accurate realistic head models based on anatomical images obtained using magnetic resonance imaging (Bentley et al. 2001; Bentley et al. 2002; Bentley et al. 2003; Iannetti et al. 2003) . These studies have repeatedly identified bilateral opercular (SII, insula) and anterior cingulate (ACC) cortical regions as significant contributors to the LEP waveform (see Garcia-Larrea et al. 2003 for a review).
Comment on source localization methods Source localization methods rely on mathematical models of the bio-electrical generators and the volume conductors within which they lie. The forward problem consists in modeling the scalp electromagnetic fields produced by a known source configuration. This modeling requires knowledge of the complex geometry and the different conduction properties of cerebral tissues constituting the brain and its envelopes. Several head models have been proposed. The earliest models consider the head as a series of concentric spheres, each layer corresponding to a different tissue whose conductivity is assumed to be homogeneous. The most used model is constituted of three layers (skin, skull, and an inner medium encompassing all intracranial content). A four-layer model which includes a thin intermediate layer modeling cerebro-spinal fluid is sometimes used. Recently, more realistic surface head models have been proposed. These models are based on the segmentation of anatomical images obtained using
31
magnetic resonance imaging (MRI). The construction of these models relies on the ability of segmentation algorithms to separate the different tissues of the head. These models offer the advantage of taking into account the complex geometry as well as the significant inter-subject variability of the conductor volume. However, as these models are difficult to implement and computationally consuming, spherical head models are still the most used models to solve the forward problem. The inverse problem consists in finding the source configuration which could explain the electromagnetic activity recorded at the scalp. The difficulty in solving the inverse problem is that an infinite number of source configurations can produce equal results at the scalp. Dipolar modeling techniques assume that the event-related electrical brain activity is concentrated in areas whose sizes are small as compared to their distance from the recording electrodes. Indeed, under these conditions, activity within these areas can be assimilated to a single equivalent dipole. Spatio-temporal dipolar-modeling algorithms, introduced by Scherg and Von Cramon (1986), assume that the position (and sometimes the orientation) of equivalent dipoles stay constant within an arbitrarily defined temporal window. The temporal sequence of activity is then explained by a variation in strength (and orientation) of these dipoles. The location, magnitude, and orientation of dipolar sources are then iteratively optimized until the scalp distribution of electromagnetic fields produced by the given configuration best coincides with the actual scalp recording. The key limitation of these methods is the necessity to predefine the number of active sources. This a priori assumption is crucial as it will determine whether a given solution actually provides neurophysiological information about where the recorded signals are generated in the brain. Recent methods have attempted to circumvent the important problem related to arbitrarily defining the number of sources. These methods are based on a ‘distributed source’ model (Dale and Sereno 1993; Pascual-Marqui et al. 1994; Baillet and Garnero 1997). Distributed source models consist in the reconstruction of the brain electrical activity in each point of a three-dimensional mesh. As each point of the mesh is considered a possible location of a current source, no a-
32
priori assumption as to the number of dipoles is required. However, as an infinite number of current source distributions can produce the exact same scalp potentials, different constraints are required to identify the most physiologically acceptable solution. The space within which sources are allowed is another important constraint defined by the investigator. For dipole-based algorithms, this constraint is defined by the space that is included in the search procedure. For distributed source algorithms, this constraint is defined by the location of the points constituting the three-dimensional solution mesh. As a-priori assumptions are required to define these constraints, the validity of a given source configuration is conditioned by the validity of these assumptions. Therefore, use of prior experimental results to constrain solutions should be used with caution as this could lead to the introduction of systematic errors or bias.
4.1.2.1
Bilateral operculo-insular cortices
Tarkka and Treede (1993) were the first to apply source analysis methods to brain responses elicited by laser stimulation. Results of that initial study proposed that bilateral activity originating from operculo-insular regions largely contributed to the observed LEP waveforms. These activities were interpreted as arising bilaterally from secondary somatosensory cortices (SII). The earliest activity was recorded contralateral to the stimulation site, peaking at 160 ms after stimulation of the hand dorsum. As compared to the contralateral activity, the ipsilateral activity was delayed, peaking at 240 ms after stimulus onset. Stimulating different body parts did not lead to significant changes in the configuration of the dipole model. Using a similar dipole-modeling technique to examine brain responses elicited by laser stimuli applied to the temple, Bromm and Chen (1995) provided additional results suggesting that bilateral operculo-insular sources participate in the generation of LEPs. There again, a slight delay between contralateral
33
(peaking at 106 ms) and ipsilateral (peaking at 112 ms) responses was observed. In accordance with the previous study, the activity was hypothesized to originate from bilateral SII cortices. The dipolar model proposed by Valeriani et al. (1996) also included two dipoles with a slightly delayed time-course in the contralateral and ipsilateral opercular regions. In this study, it was initially proposed that bilateral hippocampal activity additionally contributed to the LEP responses. However, using a method to project coordinates from a spherical head model onto Talairach space, this activity was later reinterpreted as possibly originating from bilateral insular regions (Garcia-Larrea 1998). Valeriani et al. (2000) modeled sources located in the upper bank of the sylvian fissure to explain LEPs elicited by laser stimulation of both the hand and foot (figure 1-9). This activity was interpreted as originating from SII cortices but a contribution of insular regions was not excluded. The body location of the eliciting stimulus did not modify the location of these sources. Such as in previous studies, the contralateral response (peak latency: 157 ms for hand stimulation, 217 ms for foot stimulation) preceded the ipsilateral response (peak latency: 180 ms for hand stimulation, 253 ms for foot stimulation). A peculiar finding of this study was that the delay between ipsilateral and contralateral suprasylvian responses was markedly longer for responses elicited by stimulation of the foot (36.3 ms) than for responses elicited by stimulation of the hand (22.6 ms). This difference was interpreted by investigators as either resulting from different transcallosal transfer times or resulting from the ipsilateral activity not being triggered by the contralateral activity. To explain the later part of the LEP response, two deeper dipoles were included in the model. The location of these dipoles was compatible with bilateral insular sources (Garcia-Larrea 1998). Their temporal course displayed a biphasic activation pattern (peak latencies 239 and 373 ms for hand stimulation, 319 and 459 ms for foot stimulation). Schlereth et al. (2003) also proposed that bilateral SII and insular cortical regions participate in the generation of LEPs. The magnitude of this activity, peaking at 155 ms after stimulation of the hand dorsum, was shown to be correlated with the intensity of the laser stimulus.
34
A.
B.
Fig. 1-9. Five dipole spatiotemporal solution for the grand average of laser-evoked potentials recorded to the right hand (A) and foot (B). The residual variance was 4.2% and 4% in the hand and foot models respectively. Left waveforms represent the time course of modeled sources. Peaks of activity are pointed by vertical arrows. Right head views illustrate the location and orientation of modeled dipoles. The upper rows (1 and 2) show the source potentials and locations of the dipoles in the upper bank of the Sylvian fissures. The medial dipole (3) is displayed in the middle row. The lower rows (4 and 5) show the locations of of the two mesio-temporal dipoles. Note that the location of these last two dipoles were later reinterpreted as corresponding to insular sources (Garcia-Larrea 1998). From Valeriani et al. (2000).
35
Although most studies have consistently described opercular sources after laser stimulation, some exceptions exist. Using a realistic head model, Bentley et al. (2001) modeled the responses elicited by CO2 laser stimuli applied to the right forearm of a single subject. The resulting model did not include generators in opercular regions. Indeed, the sources proposed by these investigators consisted in a contralateral posterior parietal activation, peaking at 260 ms, followed by cingulate activity, peaking at 400 ms, and anterior insular activity, peaking at 800 ms. Using a similar realistic head model, Ianetti et al. (2003) compared dipolar modeling solutions of LEPs evoked by coactivation of A - and C-fiber nociceptors with that of LEPs evoked by selective activation of C-fibers. The stimuli were applied to dorsal spinous processes. Such as the study of Bentley et al. (2001), no significant opercular activity was found to contribute to the responses elicited by stimuli co-activating A - and C-fiber nociceptors. Bilateral opercular activity was, however, hypothesized to contribute to the LEP responses elicited by selective activation of C-fibers. These contradicting results may have been related to differences in signal-tonoise ratio. Indeed, these studies only revealed sources whose latencies were compatible with the later part of the LEP response (i.e. the LEP P2 component and the later part of the LEP N2 component). Albeit these exceptions, most dipolar modeling studies have shown that bilateral operculo-insular regions may significantly contribute to the generation of LEPs (figure 1-10). These source locations are compatible with signals originating from bilateral SII and deeper insular cortices. However, the proximity of these cortical regions and the relatively low spatial resolution of these methods do not allow clearly distinguishing both cortical regions. Another consistent finding of these studies is the observation of a delay between contralateral and ipsilateral opercular activities. This delay (approximately 10 – 20 ms) is compatible with hypothesized transcallosal transfer times. Several studies have shown that changing the body location of the stimulus does not significantly modify the location or orientation of operculo-insular dipoles, suggesting the absence of clear-cut somatotopical organization of underlying cortical generators. The magnitude of this activity 36
was shown, however, to vary as a function of stimulus intensity. Most of these studies have described the contralateral opercular activity as the earliest recorded signal in response to laser stimuli. Its latency was, in most cases, similar to that of the LEP N1 component. However, the temporal patterns of SII sources suggest that they also contribute to the generation of the later N2 LEP component.
Fig. 1-10. Anatomical locations of suprasylvian laser-evoked potential sources reported in twelve studies were projected onto a 3D-MRI normalised in Talairach space. White lines cross the anterior commissure in axial and sagittal slices, and indicate the level of the ACPC line in coronal slices. The projection was performed either directly, from Talairach coordinates or by transposition of spherical coordinates using the method of Merlet et al. (1999). Although inter-study variability was important in the anterior-posterior axis, the overall distribution of sources closely followed the axis of the Sylvian sulcus. Pale blue circles and yellow triangles represent data from intracranial recordings. From Garcia-Larrea et al. (2003).
4.1.2.2
Anterior cingulate cortex
In addition to identifying bilateral opercular sources, studies (Tarkka and Treede 1993; Bromm and Chen 1995; Valeriani et al. 1996; Valeriani et al. 2000b; Bentley et al. 2002; Bentley et al. 2003; Iannetti et al. 2003) have repeatedly proposed that activity arising from locations within the anterior cingulate cortex (ACC) significantly contributes to the observed LEP
37
responses (see Figure 1-9 and 1-11). Indeed, in the pioneer study by Tarkka and Treede (1993), the proposed four dipole model included a dipole whose location was compatible with that of the ACC. After stimulation of the hand, onset of this activity was approximately 240 ms. Therefore, it was assumed that the ACC activity mostly contributed to the LEP P2 component.
Fig. 1-11. Laser-evoked anterior cingulate (ACC) sources reported in eleven different studies were projected onto Talairach space for inter-study comparison. All dipoles were projected onto the same parasagittal slice (x = 4 mm), even though their location in the x axis ranged from -1 to +8 mm. Note the relatively posterior location of many sources within the ACC, most of them lying at or between the anterior and posterior commissures. Grey circle: selective activation of C-fiber nociceptors. From Garcia-Larrea et al. (2003).
The four dipole model proposed by Bromm and Chen (1995) also included a dipole located in “deep midline brain structures”. Activity of this dipole peaked 150 – 220 ms after stimulation of the temple. These latencies were mostly coincident with the latency of the P2 component. In the dipole model proposed by Valeriani et al. (1996), a frontal dipole, very close to the midline, and possibly corresponding to the anterior cingulate gyrus was also added to explain the later part of the LEP response (see dipole 3 of Figure 1-9). The temporal course of its activity was described as biphasic. The first peak (~190 ms) was hypothesized to contribute to the earlier part of 38
the N2 component. The second peak (~290 ms) was coincident with the P2 component. The finding that the ACC dipole contributed mostly to the earlier part of the N2 component while bilateral opercular dipoles contributed most to the later part of the N2 component led these investigators to postulate that the LEP N2 component is not a unitary phenomenon but rather reflects the temporal overlapping of two distinct subcomponents (labeled ‘N2a’ and ‘N2b’). A similar biphasic ACC response was described in a study of the same group, comparing dipole configurations explaining LEPs evoked by stimulation of the hand to that evoked by stimulation of the foot (Valeriani et al. 2000). When stimulating the hand, the peak latency of both activities were respectively 217 and 333 ms. When stimulating the foot, the peak latency of both activities were respectively 281 and 406 ms. While most dipolar modeling studies have pointed to a source originating in the midportion of the anterior cingulate, corresponding to Brodman’s area 24 (BA24), some studies have proposed that this activity could originate from slightly different locations. Indeed, using a realistic head model, Bentley et al. (2001, 2003) proposed a model with a dipole located at the border between the left anterior and posterior cingulate cortex (BA23) when stimulating the right forearm. 4.1.2.3
Primary somatosensory cortex
Whether or not the primary sensory cortex (SI) participates in the recorded LEP responses remains equivocal. The four dipole model initially proposed by Tarkka and Treede (1993) included a source located in the contralateral primary sensory cortex. This activity was concomitant with that originating from bilateral SII areas. However, and unlike the other dipoles (bilateral SII, ACC), the contralateral SI dipole changed location when stimulating different body parts, suggesting a somatotopical organization of the underlying source. However, after this initial study, most studies have proposed dipolar modeling solutions of LEP responses which did not include a contralateral SI generator (Bromm and Chen 1995; Valeriani et al. 1996; Valeriani et al. 2000b; Bentley et al. 2001; Schlereth et al. 2003). It seems therefore that the bulk of recorded LEP responses may be explained without assigning a source into SI regions. 39
These observations do not necessarily mean that the laser stimulus does not generate activity within the contralateral SI cortex. Indeed, it could well be that the laser stimulus does produce activity within SI but that this activity does not significantly contribute to the generation of laser-evoked brain potentials*. 4.1.2.4
Additional sources
EEG modeling studies of LEPs have occasionally proposed the participation of other source locations. These sources have been described with much less consistency than the above-mentioned operculo-insular and cingulate responses. Valeriani et al. (1996) proposed that in addition to bilateral opercular and cingulate sources, two symmetrical sources located in mesiotemporal regions could contribute to LEP responses. It was first proposed that these two sources were located in the amygdala. However, localization of these dipoles was difficult to confirm as mesio-temporal dipoles were very close to the anterior insular cortex and the spatial resolution of the modeling technique may not be high enough to clearly differentiate between these two anatomical structures. It should be noted that due to the ‘closed-field’ geometrical configuration of the amygdala, laser-evoked activity occurring within these structures may not produce measurable electrical activity at the scalp. Using a realistic head-model to examine LEP responses in a single subject, Bentley et al. (2001) proposed that early posterior parietal generators could also contribute to the LEP negative components, occurring 150 – 250 ms after stimulation of the forearm.
*
As compared to LEPs, the amplitude of the early somatosensory N20 component elicited by
transcutaneous electrical activation of the median nerve, and reflecting activity arising from the contralateral SI area, is very low (e.g. 2.34 ± 0.62 µV; Oh 1993). Therefore, individualizing the N20 component requires repeating the stimulus at least several hundreds of times. Hence, the fact that contralateral SI regions do not significantly contribute to LEP responses is not a strong indication that the nociceptive stimulus does not elicit brain responses within these regions as in LEP recording paradigms, the stimulus is repeated a much fewer number of times (in most cases, less than fifty). Furthermore, one could expect that the afferent volley produced by the laser stimulus would be much less synchronous than that produced by the electric stimulus.
40
4.1.3 Comparing LEPs to laser-evoked responses obtained using other electrophysiological recording techniques. 4.1.3.1
Magnetoencephalography: LEPs vs. LEFs Comment on magnetoencephalography
Comparing laser-evoked responses recorded using EEG to those recorded using magnetoencephalography (MEG) could be interesting as the information recorded by both techniques may be complementary. Indeed, several reports have suggested that the information provided by EEG and MEG recordings are not redundant. It is often stated that while radial components are preponderant in EEG signals, MEG recordings are blind to this radial activity and thus selectively pick up tangential activity. In addition, deep sources are thought to contribute very little to MEG signals while superficial sources could be more accurately picked up by MEG. Indeed, as the skull has a low conductivity to electric currents but is transparent to magnetic fields, MEG recordings could be more spatially accurate than EEG recordings which are blurred by the interposed layers. However, these assertions have been questioned by a few recent studies (Malmivuo et al. 1997; Baumgartner 2004; Malmivuo and Suihko 2004). Indeed, it was proposed that when the distance between adjacent electrodes or coils is very small, the signals recorded by MEG planar gradiometers could in fact be very similar to those which are measured by EEG. In other words, the information provided by both types of recordings would be most redundant. Using a concept called ‘half-sensitivity volume’ to compare the respective spatial resolution of EEG and MEG, it was postulated that although the high resistivity of the skull decreases the spatial resolution of the EEG, it does not make it worse than that of MEG (Malmivuo et al. 1997). In fact, if special care was taken to address the considerable influence of the shape and conductivity of the volume conductor, the localization accuracy of EEG could be equivalent or even superior to that of MEG (Malmivuo and Suihko 2004). However, when simple spherical head models are used, dipole localization techniques relying on MEG recordings may be more accurate than those relying on EEG recordings.
41
Kakigi et al. (1995) examined the MEG signals elicited by CO2 laser stimuli applied to the dorsum of the hand and foot. A response was recorded 150 ms and 200 ms respectively after stimulation of the hand and foot. Using a spherical head model, the sources of these activities were located bilaterally in opercular regions hypothesized to correspond to SII. Bilateral sources compatible with the location of SII have been consistently reported by all subsequent laser-evoked magnetic fields (LEF) studies (Watanabe et al. 1998; Ploner et al. 1999; Yamasaki et al. 1999; Kanda et al. 2000; Timmermann et al. 2001; Ploner et al. 2002; Raij et al. 2003). The delay between the onsets of contralateral and ipsilateral responses found in EEG studies was also reported in several MEG studies (Ploner et al. 1999; Kanda et al. 2000; Timmermann et al. 2001; Ploner et al. 2002; Raij et al. 2003). Unlike LEP studies, most LEF studies have failed to reveal responses arising from the cingulate cortex (Kakigi et al. 1995; Watanabe et al. 1998; Yamasaki et al. 1999; Kanda et al. 2000). Several propositions have been made to explain why MEG does not pick up the cingulate activity which is recorded using EEG. The first is that the laser-evoked cingulate activity is not sufficiently synchronized to produce magnetic fields susceptible of being modeled by equivalent dipoles (Kanda et al. 2000). The second is that bilateral cingulate activation could produce a cancellation of magnetic fields. However, these two explanations are not very plausible as they do not explain why EEG recordings consistently pinpoint cingulate activity as an important contributor to the LEP waveform. Garcia-Larrea et al. (2003) proposed that the absence of cingulate activity within laser-evoked MEG responses could be related to its predominantly radial orientation, assuming that such an orientation would make it relatively undetectable by MEG sensors. As MEG responses are poorly sensitive to deep sources, the important distance between the scalp and the generators of cingulate activity may also be a plausible explanation as to why MEG most often fails to reveal these responses. An exception to these observations can be found in two recent studies (Ploner et al. 2002; Forss et al. 2005). However, the ACC activity described by Forss et al. (2005) was identified in only three out of ten 42
subjects. The.location (BA32) and latency (280 ms) of the cingulate activity described by Ploner et al. (2002) was quite unusual as compared to what has been repeatedly reported in EEG-based studies.
Fig. 1-12. Dipolar modeling of laser-evoked magnetic fields elicited by laser stimulation of the left hand. A. location of the ipsilateral SII source. B. Location of the contralateral SII source. C. Location of the contralateral SI source. Right panel displays the temporal course of dipole strength and goodness of fit. Note that the peak of activity of contralateral SI and contralateral SII sources coincided (vertical dashed line). Adapted from Kanda et al. (2000).
While MEG studies have failed to consistently identify the cingulate source described in EEG studies, they have more and more convincingly shown that the laser stimulus could evoke responses at locations compatible with the contralateral primary sensory cortex. Kanda et al. (2000) described a contralateral SI source which contributed to the LEF response to stimulation of the hand (see figure 1-12). The latency of this source was similar to that of two contralateral SII sources (~210 ms). Several other MEG source-modeling studies have confirmed that including a contralateral SI dipole allowed a better fit of LEF responses (Timmermann et al. 2001; Ploner et al. 2002; Raij et al. 2003; Forss et al. 2005). In most of these studies, the latency of the contralateral SI activity was concomitant or even slightly delayed as compared to the early contralateral SII activity.
43
Fig. 1-13. Event-related magnetic fields (MEG) and eventrelated potentials (EEG) were recorded in one subject. ES: intracutaneous electrical stimulation hypothesized to preferentially activate A -fiber afferents. TS: transcutaneous electrical stimulation preferentially activating large-diameter A -fiber afferents. The time-course of contralateral SI, insular, SII, and MT sources were obtained by dipolar modeling of MEG responses. Because the early SI response to ES appeared 72 ms later than that to TS, this figure compared responses to ES with responses to TS with a time-shift of 72 ms. Furthermore, the vertical scale for ESevoked SI activity was enlarged two times to clearly show its waveform. Arrowheads indicate the latency of the early and late peak of SI activity. Notice the similarity between ES and TS responses. From Inui et al. (2003).
In a recent study (Inui et al. 2003), using a special form of intradermal electrical stimulation to selectively activate thin cutaneous A -fiber (and possibly C-fiber) afferents, it was suggested that the time-course of the contralateral SI activity was biphasic (see Figure 1-13). Latency of the later peak was similar to that of the contralateral SII activity. However, the first peak of activity corresponded to the earliest activity recorded in this study. In other words, results suggested that the earliest activity which may be recorded using MEG in response to selective activation of A -fiber afferents is located in contralateral SI and not SII regions, as had been suggested by all previous LEP and LEF studies. To explain the uniqueness of these results, the
44
investigators proposed that, as compared to the laser stimulus, the electrical stimulus allowed a better synchronization of the A -fiber afferent volley, making it possible to identify this early SI activity. However, these results have not yet been replicated and the selectivity of the electrical stimulator may need to be better determined. 4.1.3.2
Intracranial recordings: LEPs vs. LFPs
Human intracranial recording of local-field potentials (LFPs), using either subdural or implanted electrodes, have brought direct proof as to whether or not the cortical regions pointed by source modeling studies do indeed respond to laser stimuli. These recordings are usually performed using electrodes implanted for evaluating the indication of functional surgery in drug-resistant focal epilepsy. Lenz et al. (1998a) examined responses from six subjects to laser stimulation of the hand dorsum and face using subdural electrodes placed over left frontotemporal areas. The recorded potentials consisted of a negative-positive complex. When stimulating the hand, latency of the negative peak was approximately 220 ms for contralateral stimulation and 250 ms for ipsilateral stimulation. Latency of the positive peak was approximately 380 ms for contralateral stimulation and 440 ms for ipsilateral stimulation. These responses were maximal over the parietal operculum, suggesting that their generators were located in SII and / or in the insula. The studies by Frot et al. (Frot and Mauguiere 1999a; Frot et al. 2001; Frot and Mauguiere 2003) have brought definitive proof that laser stimulation evokes responses originating from operculo-insular regions. These studies examined laser-evoked responses recorded using deep implanted electrodes whose tracts included contacts located both in SII and in the insula (figure 114). Laser stimuli were shown to evoke temporally distinct bilateral responses in SII and in the insula. The first response consisted in a negative-positive
45
wave (N140-P170) recorded at the more lateral contacts, compatible with the location of SII. The second response consisted in a negative-positive wave (N180-P230) recorded at more medial contacts, compatible with the location of the insula. The insular response, beginning approximately 180 ms after stimulus onset, was delayed as compared to the SII response, beginning approximately 140 ms after stimulus onset. Furthermore, the ipsilateral responses from both the insula and SII were delayed by approximately 15 ms as compared to the contralateral responses.
Fig. 1-14. Contralateral laser-evoked potentials were recorded in the post-rolandic operculo-insular cortex of one patient (earlobe reference recording). The operculoinsular electrode (E) is represented on the patient’s MRI slice (-5 mm caudal to the VAC plane and 8 mm above the AC-PC plane. Contacts in black are located in the insular cortex. Contacts in grey are located in the suprasylvian cortex. ML = median line; AC-PC = horizontal anterior commissure-posterior commissure plane. Two distinct negative-positive responses were recorded. The first (N140-P170) was recorded at supra-sylvian contacts. The second (N180-P230) was recorded at insular contacts. From Frot and Mauguière (2003).
46
In another study, Lenz et al. (1998b) examined laser-evoked responses recorded in five patients using subdural electrode grids placed bilaterally over the convexity and medial wall of the frontal lobe. After stimulation of the face, a biphasic response was recorded at locations compatible with the anterior cingulate cortex (BA24). The first peak of activity occurred between 211 and 243 ms. The second peak of activity occurred between 325 and 352 ms. For unknown reasons, investigators failed to elicit similar responses when stimulating the upper limbs*. Nonetheless, Garcia-Larrea et al. (2003) reported the recording of laser-evoked activity in the ACC (BA24 and possibly BA32) after stimulation of the dorsum of the hand. The latency of the activity within BA24 was consistent with that of the scalp laser-evoked vertex potentials. Kanda et al. (2000) examined responses to laser stimulation of the hand using subdural electrodes placed over the contralateral primary sensory cortex. A signal was recorded at a latency of approximately 220 ms. The spatial distribution of this activity suggested that its source was probably not located in area 3b but rather in the crown of the post-central gyrus. This hypothesis was further supported by a study from Valeriani et al. (2004). Indeed, in this study, responses to electric stimuli activating large myelinated A -fiber afferents and responses to laser stimuli selectively activating A - and C-fiber afferents were examined using an electrode located in area 3b of the primary sensory cortex. Although approximately 20 ms after stimulus onset, a reliable component was recorded in response to the electrical stimulus, no response could be recorded in response to the laser stimulus. 4.2
The C-fiber mediated ultra-late LEP
4.2.1 Latency, morphology, and topography Bromm et al. (1983) were the first to report that a laser stimulus which activates C-fiber nociceptors selectively could elicit reproducible ultra-late *
A possible explanation to this discrepancy is that it was related to the more intrusive or
salient quality of stimuli applied to the face as compared to that applied to the upper limbs (see Section 5.3.2 of Introduction).
47
event-related brain potentials. In that initial study, selective activation of Cfibers was obtained by applying a prolonged pressure to the superficial radial nerve, thereby producing an ischemic block selectively affecting myelinated Afibers. The LEP response was described as an ultra-late positivity, maximally recorded at the vertex. The latency of this component was approximately 1260 ms. Bromm and Treede (1987b) replicated this finding and also showed that ultralate C-fiber related LEPs could be recorded in patients suffering from pathologies selectively affecting myelinated nerve fibers. The ultra-late response was described as a negative-positive complex, maximal at the vertex, and occurring within 900 – 1500 ms after stimulus onset. It should be noted that in these earlier studies, ultra-late responses were revealed using latency correction algorithms. The rationale behind the use of such algorithms was that the C-fiber afferent volley would be poorly synchronized, due to the greatly varying conduction velocity of these slow afferent fibers. However, studies using such latency-correction algorithms should be interpreted with caution as the use of these techniques may lead to the enhancement of artifactual potentials (Treede and Bromm 1988). In addition to blocking the A -fiber afferent volley by applying an ischemic pressure block preferentially affecting myelinated A-fibers (Bromm et al. 1983; Bromm and Treede 1987b; Nahra and Plaghki 2003b), two other methods which allow to selectively activate C-fiber nociceptors have been proposed (see Plaghki and Mouraux 2002 for a review of these methods). The first method is based on the lower thermal activation threshold of C-fiber nociceptors as compared to that of A -fiber nociceptors. The method consists in lowering the intensity of the laser stimulus such that the skin temperature reaches the threshold of C-fiber nociceptors but not that of A -fiber nociceptors. This method has been used successfully by several studies (Towell et al. 1996; Magerl et al. 1999; Valeriani et al. 2002a; Cruccu et al. 2003; Iannetti et al. 2003). The second method is based on the denser skin distribution of C-fiber nociceptors as compared to that of A -fiber nociceptors (Ochoa and Mair 48
1969; Schmidt et al. 1994). The method consists in using a very small stimulus surface area (e.g. 0.15 mm2; Bragard et al. 1996). Indeed, under these conditions, there is a high probability that C-fiber nociceptors will be activated
by
the
stimulus
without
concomitantly
activating
A -fiber
nociceptors. Numerous studies (Bragard et al. 1996; Opsommer et al. 1999; Opsommer et al. 2001a; Opsommer et al. 2001b; Qiu et al. 2002; Opsommer et al. 2003) have successfully used this method which offers several advantages as it is non invasive, highly reproducible, and does not require lowering the energy density of the stimulus.
(µV)
Fig. 1-15. Selective activation of C-nociceptors was obtained by using tiny skin surface areas (~0.23 mm2) laser stimuli. Grand-average of ultra-late laser-evoked potentials (mean ± SE) recorded at electrode CZ from nine subjects. Stimuli were applied to the dorsum of the left hand. Vertical dashed lines represent the latency of ultra-late LEP peaks. Scalp maps (a,b, and c) display the topographical distribution of the signal at these three latencies (adapted from Opsommer et al., 2001).
49
Whatever the method used, all of these studies have confirmed that when the activation of A -fiber nociceptors is avoided, or when the peripheral A -fiber afferent volley is blocked, selective activation of C-fibers elicits an ultra-late negative-positive complex whose latency (750 – 1150 ms after stimulation of the hand) is compatible with the conduction time of C-fibers (for an example, see panel B of figure 1-5).
Fig. 1-16. Grand-average traces of ultra-late laser-evoked potentials recorded at PZ, T4, CZ, T3, and FZ electrodes after selective activation of C-fiber nociceptors using non-painful, low intensity, CO2 laser stimuli applied to the right perioral region in neutral, distraction, and attention conditions. The ultra-late N1 response appeared as a shoulder on the rising branch of the ultra-late N2 potential. The ultra-late N1 component was best individualized at the contralateral T3 electrode rereferenced to FZ. Adapted from Valeriani et al. (2002).
The morphology and topography of these ultra-late responses (see figure 115) very much resembles that of the late LEP response elicited by A -fibers (see figure 1-5). The most prominent components of both responses consist of a negative-positive complex, referred to as the ultra-late N2-P2. The scalp distribution of late and ultra-late N2-P2 complexes are very similar 50
(Opsommer et al. 2001a; Valeriani et al. 2002a; Forss et al. 2005). Such as the late N2, the ultra-late N2 is maximal at the vertex but tends to spread bilaterally onto temporal sites. Such as the late P2, the ultra-late P2 displays a widespread central topography, maximal at the vertex. Valeriani et al. (2002) reported than in addition to evoking ultra-late N2 and P2 components, selective activation of C-fibers could also elicit an earlier negative component, labeled ‘ultra-late N1’ (see figure 1-16). The ultra-late N1 was very similar to the N1 component which precedes the late N2-P2 complex in response to A nociceptor stimulation. Indeed, such as the late N1, the ultra-late N1 was described as occurring during the “ascending shoulder” of the ultra-late N2 and, more importantly, displayed a similar contralateral temporal topography. Therefore, it appears that the components of the ultra-late LEP response elicited by selective activation of C-fibers share a common morphology and topography with the late LEP components elicited by the activation of A fibers. 4.2.2 Hypothesized cortical generators A few studies (Opsommer et al. 2001b; Cruccu et al. 2003; Iannetti et al. 2003) have applied source analysis methods to ultra-late LEP responses elicited by selective activation of C-fiber nociceptors (for an example, see figure 1-17). Using 64 channel EEG recordings and a realistic head model, Opsommer et al. (2001b) proposed that a three dipole model could explain the ultra-late LEP response.
This
model
included
two
symmetrical
opercular
dipoles,
hypothesized to reflect activity originating from SII and one dipole in the median region, hypothesized to reflect activity originating from the ACC. A similar dipole model was proposed by Ianetti et al. (2003). In a study examining responses to trigeminal stimuli selectively activating Cfibers, Cruccu et al. (2003) proposed a dipolar model which also included bilateral opercular and anterior cingulate sources. In addition to these sources, this study proposed that bilateral insular regions contributed to the 51
recorded ultra-late LEPs. It should be noted that the location of the cingulate source was more anterior (BA32) than what is usually reported in studies examining LEPs elicited by A -fiber afferents (BA24). However, some studies have suggested that A -fibers could produce activity within these more anterior regions (Ploner et al. 2002).
Fig. 1-17. Source analysis of ultra-late laser-evoked potentials elicited by selective activation of C-fiber nociceptors (obtained using low intensity stimuli applied to the skin at vertebral spinous processes). A 3-dipole model gave the best source explanation. One dipole (A) was located in the posterior part of the anterior cingulate cortex. Two dipoles (B and C) were bilaterally located in operculo-insular cortices. Dipoles are overlayed on individual magnetic resonance images. A. Sagittal section through the cingular dipole. B and C. Axial and coronal sections through one of the operculoinsular sources. From Iannetti et al. (2003).
Therefore, these different source analysis studies have shown that ultra-late LEPs elicited by selective activation of C-nociceptors could be explained by dipole configurations similar to those explaining late LEPs elicited by the activation of A -nociceptors. A few studies have examined ultra-late MEG responses elicited by selective activation of C-fibers (Tran et al. 2002; Kakigi et al. 2003; Qiu et al. 2004; Forss et al. 2005). These studies similarly proposed a bilateral activation of opercular regions. Furthermore, with the exception of the study by Forss et al. (2005), all these studies included a dipole whose location was compatible with activity originating from the contralateral SI. Finally, it should be noted that three of these studies (Kakigi et al. 2003; Qiu et al. 2004; Forss et al. 2005) identified a cingulate generator. These findings are very similar to that of late LEF responses elicited by A -fibers.
52
5 The ‘common generators’ hypothesis 5.1
Late and ultra-late LEP responses may share common generators
The similarities between A - and C-fiber related late and ultra-late LEPs has led a number of investigators to propose that both responses reflect the activation of common generators (Bromm and Treede 1987b; Treede et al. 1988b; Magerl et al. 1999; Plaghki and Mouraux 2002; Kakigi et al. 2003). Indeed, the morphology and topography of both responses are very similar. Both responses are composed of a prominent negative-positive complex, recorded maximally over the vertex. Both responses also evoke an earlier and lateralized negativity, recorded maximally at contralateral temporal leads. In addition, source analysis studies have shown that similar dipole configurations could produce both late and ultra-late LEP responses. Indeed, both late and ultra-late LEPs may be explained by bilateral sources in operculo-insular regions and a median source in the anterior-cingulate. Furthermore, source analyses of MEG recordings have suggested that an additional source, probably originating from the contralateral SI cortex, contributed to both late and ultra-late LEF waveforms. While these observations constitute converging evidence indicating that late and ultra-late LEPs reflect cortical activity common to the processing of both afferent volleys (i.e. that the very same neural populations respond to both A - and C-fiber input), two alternative hypotheses should be considered. The first is that late and ultra-late responses reflect activity originating from distinct yet spatially intermingled neural populations. The second is that late and ultralate LEP responses reflect spatially separate activities but that this difference cannot be resolved using scalp EEG recordings. 5.2
Vertex potentials
Vertex potentials elicited by auditory stimuli were initially described by Davis (1939) in the raw unaveraged EEG. A similar vertex component, evoked by somatosensory stimuli, was also described in early EEG recordings (e.g. Bancaud et al. 1953). In fact, it appears that vertex potentials may be elicited 53
by sensory stimuli regardless of their modality. Indeed, vertex potentials have been described in the auditory (reviewed in Naatanen and Picton 1987), the somatosensory (Desmedt and Robertson 1977; Goff et al. 1977; Josiassen et al. 1982; Michie et al. 1987; Desmedt and Tomberg 1989; Garcia-Larrea et al. 1991; Garcia-Larrea et al. 1995), and the visual modalities (Simson et al. 1976; Simson et al. 1977; Kenemans et al. 1993; Makeig et al. 1999; Hopf et al. 2000; Potts and Tucker 2001; Potts 2004; Potts et al. 2004). Several investigators (Treede et al. 1988a; Kunde and Treede 1993; Garcia-Larrea et al. 1997) have pointed out that the N2-P2 complex of LEPs might be related to these vertex potentials as they share many similarities (see figure 1-18). Kunde and Treede (1993) compared the latency, morphology, and topography of LEPs to that of somatosensory evoked-potentials (SEPs) elicited by transcutaneous electrical stimulation of large A -fibers. Results showed that the topography and morphology of the N2-P2 complex evoked by the laser stimulus was very similar to that of the SEP N140-P250 vertex potential.
Fig. 1-18. Grand-average of evoked potentials (C3, C4, and CZ vs. A1A2) recorded in 24 subjects (40 stimulus repetitions). SEPn: transcutaneous electrical nerve stimulation. SEPc: CO2 laser cutaneous heat stimulation. Stimuli were applied to the right hand, left hand, and left foot. Stimulus intensities were 1.5-fold threshold (solid line) and 0.75-fold threshold (dashed line). Adapted from Treede et al. (1988).
54
5.2.1 Vertex potentials in the auditory modality Late auditory vertex potentials consist of a negative component (N1) occurring approximately 75 – 150 ms after stimulus onset, followed by a positive component (P2), occurring with an approximate latency of 150 – 250 ms. The auditory N1 component appears to be composed of several anatomically and functionally distinct subcomponents. Naatanen and Picton (1987) identified three subcomponents. The first component would display a fronto-central predominance, have a peak latency of 100 ms, and be modeled by bilateral and vertically oriented sources hypothesized to be located in the supra-temporal plane. The second component would display a biphasic activation pattern with a positive peak occurring at approximately 100 ms and a negative peak occurring at approximately 150 ms. This second component would be maximally recorded over temporal areas and was therefore hypothesized to originate from radially oriented dipoles located in the lateral aspect of the superior temporal gyrus (secondary auditory cortices). The third component would consist in a negative wave occurring approximately 100 ms after stimulus onset. This last component, maximally recorded over the vertex, was interpreted as reflecting a widespread transient arousal subsequently facilitating stimulus detection, analysis, and response generation (Naatanen and Picton 1987; Picton et al. 1999). This component was modeled by radially oriented dipoles located in the frontal cortex. Giard et al. (1994) performed a source analysis study of the auditory N1 component. Results of that study suggested that the auditory N1 originated bilaterally from the supratemporal plane of the auditory cortex but also from bilateral frontal regions hypothesized to be located either in cingulate or in supplementary motor areas. The auditory P2 component was not as extensively studied (for a review see Crowley and Colrain 2004). Early studies have considered the auditory P2 to be generated mainly in the vicinity of the auditory cortex, within the temporal lobe (Elberling et al. 1980; Hari et al. 1980; Perrault and Picton 1984).
55
However, dipolar modeling studies do not allow fitting a single source solution (Jacobson 1994; Verkindt et al. 1994). Results from MEG and depth EEG recordings (Godey et al. 2001) have suggested that the generators of the auditory P2 are located in the planum temporale as well as in BA22 (auditory association complex). However, other studies have speculated that the P2 component may also receive contributions from cortical areas in the upper lip of the sylvian fissure, at or near SII (Hari et al. 1990). Whatsoever, it seems that such as the auditory N1, the auditory P2 arises from multiple sources. The location of these sources would be centered around parietal and temporal opercular regions. 5.2.2 Vertex potentials in the somatosensory modality Vertex potentials elicited by somatosensory stimuli consist of a negativity (often referred to as N1 or N140) followed by a positivity (often referred to as P2 or P250). Unlike the auditory N1 which displays a midline maximum whatever the stimulated ear, the topography of the somatosensory N1 is highly dependent of stimulus location and displays a contralateral to stimulus predominance (Bruyant et al. 1993; Garcia-Larrea et al. 1995). Garcia-Larrea (1995) proposed that such as the auditory N1, the somatosensory N1 is composed of at least two distinct subcomponents: an earlier component (labeled N120 or ‘early N1’) and a later component (labeled N140 or ‘late N1”). The N120 component, displaying a contralateral temporal predominance, was hypothesized to be generated by bilateral SII sources and reflect modality-specific sensory processes. The N140 component, displaying a symmetrical scalp topography maximal at the vertex, was hypothesized to reflect more endogenous and supramodal processes (see figure 1-19). In a study combining intracranial and scalp recordings, Allison et al. (1992) also described two distinct negativities occurring in the 100 – 150 ms range. These two activities shared topographical and functional properties with the early and late N1 components described by Garcia-Larrea (1995). Indeed, 56
scalp recordings displayed an early and lateralized N120 component whose intracranial counterpart was hypothesized to be an N100 recorded in suprasylvian regions. This early activity was differentiated from a subsequent N140 component, recorded over both hemispheres, and hypothesized to correspond to a true ‘vertex negativity’.
Fig. 1-19. Grand-average somatosensory evoked potentials recorded from 15 subjects. Standard right-hand electrical stimuli were applied during three different experimental conditions: neutral, unattended hand, and attended hand. Scalp maps illustrate the N1 distribution within the early (100-130 ms), and late (130-160 ms) analysis window. Negativity is up for all traces. In the neutral condition, a small negative wave (N120) was observed at temporal and central electrodes only (T3, C3), the distribution of which was restricted to lateral scalp sites opposite to the stimulus The topographical distribution of the scalp negativity expanded gradually in the unattended condition and reached full bilateral and symmetrical aspect during the attended condition. From Garcia-Larrea et al. (1995).
57
5.2.3 Common processes underlying vertex potentials. Vertex potentials appear to be elicited by stimuli whatever their sensory modality. As the topography of the N1 vertex potential varies across different sensory modalities, it could be considered that the N1 component reflects distinct processes, specific to each eliciting modality. However, studies within these different sensory modalities have indicated that the N1 cannot be reduced to a single component but rather that it reflects the activation of multiple subcomponents. Therefore, it could be hypothesized that some but not all of these subcomponents reflect more modality-specific processes. These could include the somatosensory N120 described by Garcia-Larrea et al. (1995) but also the first and second subcomponents of the auditory N1 as defined by Naatanen and Picton, (1987). Indeed, other later components such as the somatosensory N140 and the third subcomponent of the auditory N1 share similar scalp distributions and have therefore been proposed to reflect non modality-specific or supramodal processes. As suggested by Picton et al. (1999), these later processes could be related to exogenously-triggered orienting responses (see section 6.3.4 of Introduction). Such as the later part of the N1, it is probable that that,the P2 vertex potential reflects activities common to the processing of all sensory modalities. Indeed, the topography of the P2 appears to be similar across different sensory modalities. This relatively tardive component has been hypothesized to reflect more integrative and cognitive aspects of sensory processing. 5.3
Nociceptive specificity of laser-evoked potentials
Based on these previous observations, one should consider the possibility that at least part of the LEP N2 and P2 components reflect cortical activity common to the processing of all sensory modalities. 5.3.1 Nociceptive specificity of the laser-evoked N2 potential Anterior cingulate and operculo-insular cortices have repeatedly been identified as sources significantly contributing to the laser-evoked N2
58
potential. How these different activities may relate to the vertex negativity elicited by other sensory modalities will now be reviewed. Fronto-medial sources, probably originating from the ACC, have been proposed to contribute to the LEP N2 component, the somatosensory N1 component (e.g. the N140 subcomponent as defined by Garcia-Larrea et al. 1995), but also the auditory N1 component (e.g. the third subcomponent of the auditory N1 as defined by Naatanen and Picton 1987). In most studies, these activities have been interpreted as reflecting non modality-specific processes related to stimulus-triggered orienting responses. Furthermore, all vertex negativities have been hypothesized to receive significant contributions from signals arising around bilateral opercular regions (Allison et al. 1992; Giard et al. 1994; Garcia-Larrea et al. 1995). Bilateral SII sources have been proposed to strongly participate in the generation of the vertex negativities elicited by both laser and electric somatosensory stimulation. Therefore, it could be proposed that this activity is involved in the processing of both nociceptive and non-nociceptive somatosensory input. Unlike the somatosensory N1 component, bilateral opercular sources of the auditory N1 component have, for the most part, been ascribed to activity originating from the supra-temporal plane and the superior temporal gyrus. However, the significance of these differences should be taken with caution as the low spatial resolution of EEG and MEG recordings limit their interpretability. Several studies have proposed that in addition to SII, deeper insular regions significantly contribute to the laser-evoked potential waveform (reviewed in Garcia-Larrea 2003). As shown by intracortical recordings (Frot and Mauguiere 2003), laser stimuli indeed elicit responses within locations compatible with that of the anterior portion of the insular cortex. Laser stimuli also appear to elicit smaller responses within the posterior portion of the insula. A participation of the insula to the electrically-evoked somatosensory N1 component has also been demonstrated (Frot and Mauguiere 1999b). However, and unlike the electrophysiological activity elicited by laser stimuli, 59
intracortical recordings showed that this activity was confined to the posterior portion of the insula. A similar observation was reported in a MEG source analysis study (Inui et al. 2003) comparing the activity elicited by intracutaneous electrical stimuli selectively activating A -fiber afferents to that elicited by transcutaneous electrical activation of large A -fibers. Indeed, results of that study suggested that nociceptive A -fiber activation and nonnociceptive A -fiber activation evoked similar responses in SI, SII, and anterior cingulate regions. Both types of stimulation also elicited activity hypothesized to originate from the insula. However, the insular response to nociceptive stimuli was more anterior than the insular response to nonnociceptive stimuli. Altogether, these results would suggest that the laserevoked activity originating from the anterior insula reflects processes specifically related to the nociceptive nature of the evoking stimulus while the activity originating from the posterior insula reflects less modality-specific processes. Of note, Valeriani et al. (1996, 2000a) had proposed that the LEP N2 component consisted of the overlap of two distinct subcomponents labeled N2a and N2b. Unlike the N2a which received contributions from SII and the anterior cingulate, it was hypothesized that the LEP N2b subcomponent mostly reflected activity arising from bilateral insular or mesio-temporal regions. Therefore, it may be envisaged that the N2b subcomponent reflects processes specific of the ‘nociceptive’ modality while the earlier N2a subcomponent mostly reflect processes shared with the non-nociceptive somatosensory N1 component. 5.3.2 Nociceptive specificity of the laser-evoked P2 potential The topography of the vertex P2 positivity appears to be most invariant across auditory, visual, and somatosensory modalities. For these reasons, it could well be that the LEP vertex positivity reflects processes common to all sensory modalities. However, the fact that cingulate sources have been shown to contribute to a great part of the LEP P2 component raises the question as to whether it may be completely assimilated to the late vertex positivities described in other sensory modalities. Indeed, most studies have suggested that the auditory P2 is produced by multiple and bilateral opercular sources but not by cingulate sources (reviewed in Crowley and Colrain 2004). To 60
explain this discrepancy, the possibility that additional electrophysiological components partially overlap the laser-evoked P2 component should be considered. Indeed, in a recent set of studies, Legrain et al. (2002; 2003a; 2003b) showed that when occurrence of the laser stimulus is unexpected, part of the signal within the latency range of the laser-evoked P2 could be explained by an additional component, labeled ‘P400 effect’. This overlapping positivity was hypothesized to be related to the P3a component which has been reported in other sensory modalities (Courchesne et al. 1975; Squires et al. 1975; Yamaguchi and Knight 1991; Escera et al. 1998; Katayama and Polich 1998). The P3a has been interpreted as reflecting processes related to involuntary reorientations of attention triggered by salient and unexpected exogenous events (see also section 6.3.4 of Introduction). Several studies have indicated that the P3a component may originate from frontal regions near the anterior cingulate cortex (Baudena et al. 1995; Dien et al. 2003). Therefore, one could hypothesize that the ACC contribution to the LEP P2 component may be mostly related to the overlapping of a ‘P400 effect’. Furthermore, such an interpretation of the laser-evoked P2 might help explain why, using subdural grids placed over the medial wall of the frontal lobe, Lenz et al (1998b) recorded consistent cingulate activity when stimulating the face but not when stimulating the upper limb. Indeed, as the magnitude of the P3a is strongly modulated by the saliency of the evoking stimulus, the differences observed between both stimulation sites might in fact be related to differences in saliency. Indeed, one might expect that stimuli applied to the face would be perceived as more intrusive or threatening than that applied to the hand. 5.3.3 Nociceptive specificity of the laser-evoked N1 potential Whether the earlier laser-evoked N1 component reflects nociceptive-specific processes has been challenged by several observations. Indeed, electrical activation of large myelinated A -fibers evokes several negative ERP components which precede the vertex negativity and display a contralateral temporal topography similar to that of the LEP N1 component. Kunde and Treede (1993) proposed that the laser-evoked N1 component could be related to the SEP N110 component elicited after electrical 61
stimulation of the superficial branch of the radial nerve. Indeed, the SEP N110 component displayed a similar contralateral temporal topography, was also best identified using a frontal reference, and was hypothesized to reflect activity originating from bilateral SII regions. However, the interpeak latency between the SEP N110 and the SEP N140 vertex negativity was much shorter than the interpeak latency between the LEP N170 and N240 components. This difference was interpreted as possibly reflecting a slower central processing of A -fiber input as compared to A -fibers. Yet, different processing speeds would entail that the central processes triggered by A fibers and reflected by LEPs are different than those triggered by A -fibers and reflected by SEPs.
Fig. 1-20. Somatosensory evoked potentials elicited by electrical stimulation of the median nerve were recorded in the contralateral post-rolandic operculum. A. Earlobe reference recordings show that, in this patient, the N60-P90 potential was recorded by a single contact. B. In bipolar recordings, a polarity reversal was observed for the N60 and the P90 potential. VCA: Vertical anterior commissure plane (frontal plane). VCP: Vertical posterior commissure plane (frontal plane). AC-PC. Anterior commissure -posterior commissure (horizontal plane). From Frot and Mauguière (1999).
62
Treede et al. (1988a) had initially postulated that the LEP N1 component was related to a SEP negativity, occurring approximately 60 – 70 ms after transcutaneous electrical stimulation of the median nerve (see figure 1-18). This hypothesis was based on the fact that both responses had similar interpeak latencies to the subsequent vertex negativity. However, whereas the median nerve N60 was maximal at electrode locations C3 or C4, the laserevoked N170 was maximal at electrode locations T3 or T4. As this topographical difference suggested that both responses originated from different cortical regions, they later dismissed this hypothesis (Kunde and Treede 1993). Initial studies had proposed that the SEP N60 component originates from area 1 of the primary sensory cortex (Allison et al. 1992; Srisaan et al. 1996). However, several studies had shown the possibility of a dissociated loss of this response in lobe lesions sparing the SI area (Mauguiere et al. 1983; Stohr et al. 1983). Furthermore, using intracranial electrodes implanted in the parietal operculum, Frot and Mauguière (1999b) showed that an N60-P90 component could be recorded from both the contralateral and ipsilateral parietal operculum (see figure 1-20). The ipsilateral response was delayed by approximately 12 – 16 ms. Apart from their earlier latency, this response pattern was therefore highly reminiscent of the parieto-opercular response elicited by laser stimuli (Frot and Mauguière 1999a; Frot et al. 2001; Frot and Mauguière 2003; see also figure 1-14). Furthermore, Barba et al. (2002) concurrently examined scalp EEG recordings and local-field potentials from electrodes implanted in the parietal operculum. Results of that study indicated that the N60-N70 ERP component could be dissociated into two subcomponents: a fronto-central component (N60), possibly generated in both the SMA and SI and a temporal component (N70) generated in bilateral SII. Therefore, there are several reasons to consider that the laser-evoked N1 and the later part of the electrically-evoked N60-N70 negativity share common sources. The difference in topography between both scalp responses would be related to the fact that the SEP N60-N70 component is composed of the superposition of a fronto-central N60 component and a contralateral temporal N70 component. The interpeak latency between the SEP N60-N70 and the 63
SEP vertex negativity is similar to the interpeak latency of the LEP N1 and N2 components. Furthermore, the interhemispheric delays between contralateral and ipsilateral responses are similar. 6 Experimental modulation of late LEPs, ultra-late LEPs, and vertex potentials To further explore the possible relationships between A -fiber late LEPs, Cfiber ultra-late LEPs, and vertex potentials in general, it will now be examined how different experimental manipulations modulate these electrophysiological responses. 6.1
Stimulus intensity – Intensity of perception
6.1.1 Late LEPs Numerous studies (Carmon et al. 1976; Bromm and Treede 1984; Kakigi et al. 1989; Plaghki et al. 1994; Svensson et al. 1997; Timmermann et al. 2001; Nahra and Plaghki 2003b; Schlereth et al. 2003) have shown a positive relationship between amplitude of the A -fiber mediated late LEP P2 component and the intensity of the evoking stimulus (see figure 1-5). These studies have also shown that increasing the intensity of the stimulus could reduce the latency of LEP components. However, studies examining intensity of perception and magnitude of late LEP responses under different attentional settings (Plaghki et al. 1994; GarciaLarrea et al. 1997) have shown that the amplitude of late LEP responses may be more directly correlated with the subjective pain sensation than with the actual stimulus intensity (figure 1-21). 6.1.2 Ultra-late LEPs The amplitude of ultra-late LEPs, unlike that of late LEPs, is not significantly correlated with stimulus intensity (Bragard et al. 1996; Nahra and Plaghki 2003b). However, when brief laser stimuli are used to selectively activate Cfibers, the correlation between the subjective intensity of perception and the intensity of the laser stimulus is also very weak. Therefore, it is probable that the absence of clear correlation between stimulus intensity and ultra-late LEP 64
amplitude is related to the fact that the amplitude of ultra-late LEPs, such as that of late LEPs, is more directly related to the subjective intensity of perception.
distraction attention
Fig. 1-21. Grand-average waveforms and scalp maps of laser-evoked potentials (LEPs) from eight out of fifteen subjects having the greatest difference in subjective perception between attentive (thick line) and distractive (thin line) conditions. In the attentive condition, subjects were asked to report the number of perceived laser stimuli. In the distractive condition, subjects were to disregard the laser pulses and count occasional interruptions occurring in a 60 dB noise delivered through headphones. Scalp maps depict the voltage distribution at the three latencies (160, 220, and 350 ms) indicated by vertical lines and corresponding respectively to the laserevoked N1, N2, and P2 components. Average intensity of perception, estimated using a visual-analogue scale (VAS) was 4.2 in the attentive condition and 2.7 in the distraction condition. The enhancement of intensity of perception in the attentive condition was correlated with an enhancement of the vertex LEP response (from Garcia-Larrea et al., 1997).
65
6.1.3 Somatosensory and auditory vertex potentials It is well known that the amplitude of auditory and somatosensory vertex potentials is highly correlated with the intensity of the evoking stimulus. The influence of loudness of auditory stimuli on the N1-P2 peak-to-peak amplitude is a very consistent finding (Rapin et al. 1966; Beagley and Knight 1967; Picton et al. 1970; Gerin et al. 1972). The consensus is that this increase appears to be roughly linear in microvolts per decibel, with a tendency to saturate or even reverse at high levels. Furthermore, such as the latency of LEP components, the latencies of auditory N1 and P2 components have been reported to decrease with increasing stimulus intensity (Rapin et al. 1966; Beagley and Knight 1967). 6.2
Stimulus repetition – Interstimulus interval
6.2.1 LEPs It is widely accepted that even when care is taken to shift stimulus location, thereby avoiding habituation or sensitization of nociceptors, repeating the laser stimulus may induce an important decrease of late LEP amplitude (Bromm and Treede 1987a; Raij et al. 2003; Truini et al. 2004). Indeed, Bromm and Treede (1987a) reported that when two laser stimuli were applied with an inter-stimulus interval (ISI) of 900 ms, the amplitude of the LEP evoked by the second stimulus was significantly reduced. Raij et al. (2003) examined EEG and MEG responses evoked by trains of laser stimuli, using ISIs ranging from 0.5 to 16 seconds. To reduce stimulus expectancy, a 20% variation of ISI was introduced from trial to trial. This study showed that repetition induced an important attenuation of both LEP and LEF components. The observed amplitude decrease was described as an exponential function of the ISI. Indeed, the response recovered rapidly for ISIs ranging from 0.5 to 4 seconds and then saturated for ISIs above 8 seconds. Truini et al. (2004) examined LEP responses to pairs of laser stimuli applied to the dorsum of the hand (figure 1-22). Pairs were presented in blocks using a constant ISI which ranged from 0.25 to 2 seconds. As compared to the LEP response elicited by the first stimulus, the amplitude of the second LEP response was attenuated. At the smallest ISI (i.e. 0.25 s), the LEP response 66
was attenuated by 50%. At larger ISIs, the response gradually recovered but an amplitude decrease of 20% was still observed at 1000ms.
A.
B.
Fig. 1-22. Two consecutive CO2 laser stimuli were directed to two adjacent skin spots of the hand dorsum. After each stimulus pair, target of both stimulators were slightly shifted to adjacent spots. A. Waveforms represent the average of ten LEP trials in one representative subject. 1: control response (one single laser stimulus). 2-7. paired laser stimuli presented with interstimulus intervals of 250, 500, 750, 1000, 1500, and 2000 ms. Vertical calibration: 10 mV. Horizontal calibration: 200 ms (traces 1-6) and 500 ms (trace 7). Dashed vertical line indicates the onset of the conditioning stimulus. Vertical arrows indicate the onset of the test stimulus. B. LEP P2 recovery curve. Amplitude of the LEP P2 component was expressed as a percentage of the amplitude of the unconditioned P2 component. (mean ± SE, n = 10). X-axis: interstimulus interval (ms). Adapted from Truini et al. (2004).
No studies have examined the effect of stimulus repetition and interstimulus interval on the ultra-late LEP evoked by selective activation of C-fibers. 6.2.2 Somatosensory and auditory vertex potentials The effect of stimulus repetition and inter-stimulus interval on the latency and amplitude of vertex potentials has been extensively studied in the auditory and somatosensory modalities (Ritter et al. 1968; Roth and Kopell 1969;
67
Fruhstorfer et al. 1970; Weber 1970; Fruhstorfer 1971; Ohman and Lader 1972; Prosser et al. 1981; Angel et al. 1985; Woods and Elmasian 1986; Bourbon et al. 1987; Tomberg et al. 1989; Barry et al. 1992). Vertex potentials elicited by the first stimulus in a train are usually large in amplitude. When a constant presentation rate is used, their amplitude then rapidly diminishes with repetition, reaching a low asymptotic level after just a few stimulus presentations (for a review, see Naatanen and Picton 1987). The amplitude decrement is faster and more pronounced when short and constant ISIs are used (Fruhstorfer et al. 1970; Angel et al. 1985). For example, Tomberg et al. (1989) showed that the somatosensory N1 vertex potential could disappear when ISI is reduced to 1.4 seconds. A relatively stable inverse relationship between N1 amplitude and ISI duration, following an exponential function, has often been described (Davis et al. 1966; Zerlin and Davis 1967; Nelson and Lassman 1968; Woods et al. 1980; Woods and Courchesne 1986). According to some studies, the full recovery of the auditory N1 vertex potential could require up to 10 seconds (Davis et al. 1966; Ritter et al. 1968; Fruhstorfer et al. 1970; Naatanen 1988). However, when variable rates of stimulation are used, studies have shown that the auditory vertex potential is not necessarily attenuated by repetition and may even be enhanced when sounds are presented within intervals shorter than 400 ms (Loveless et al. 1989; Budd and Michie 1994; Loveless et al. 1996; McEvoy et al. 1997; Sable et al. 2003). To explain the effect of stimulus repetition and ISI on the magnitude of vertex potentials, several hypotheses have been put forward*. 1. Sensory adaptation or receptor fatigue Several experimental observations indicate that the decrement of somatosensory vertex potentials cannot be a mere consequence of *
It should be noted that, in a number of experimental paradigms, additional factors such as
progressive reduction of arousal may have contributed to the observed response decrements.
68
sensory adaptation or receptor fatigue. Indeed, unlike the later vertex potential components, it has been shown that the first electrophysiological response to electrical somatosensory stimulation (i.e. the N20 component) was fully recovered at ISIs as short as 200 ms (Huttunen and Homberg 1991; McLaughlin and Kelly 1993). Furthermore, stimulus repetition does not induce an attenuation of subjective intensity of perception similar to the amplitude decrement of vertex potentials. More importantly, sensory adaptation or receptor fatigue cannot be taken into account when examining the effect of stimulus repetition on LEPs as in these studies, special care is always taken to displace the location of the stimulus target. 2. Refractoriness When producing a first response, the neural populations that generate vertex potentials or LEPs could enter a transient state of ‘refractoriness’. The ability of these neurons to produce additional responses would be diminished and only gradually recover over time. The amplitude of a given response would therefore be directly related to the delay between the two response-eliciting stimuli. At high repetition rates, one would thus expect the response to the second stimulus to be minimal. This hypothesis was derived from the fact that following an action potential, single nerve cells display a ‘refractory period’. This hypothesis assumes that the polysynaptic neural assemblies generating vertex potentials or LEPs show a phenomenon similar to the ’refractoriness’ of single nerve cells. In other words, the processes underlying vertex potentials and LEPs would be subject to temporal limitations (Naatanen and Picton 1987; Budd et al. 1998). However, as complete recovery of the vertex potential amplitude may require up to ten seconds (Davis et al. 1966; Ritter et al. 1968; Fruhstorfer et al. 1970; Naatanen 1988), it seems difficult to envisage that refractoriness of simple cellular mechanisms could fully account for the response decrements induced by repetition (Naatanen and Picton 1987).
69
3. Habituation The enhancement of vertex potentials elicited by the first stimulus in a train has often been hypothesized to be related to an initial orienting-response (Kenemans et al. 1989). This stimulus would catch attention and therefore elicit a large vertex potential (Squires et al. 1975; Snyder and Hillyard 1976; Alho et al. 1998; Escera et al. 1998). The response decrement of vertex potentials and LEPs induced by stimulus repetition would thus result from a progressive loss of novelty associated with the repetition of the stimulus. In other words, the great reduction of unexpectedness or signal value of the stimulus after a first presentation would contribute to the observed ERP amplitude decrement. The fact that stimulus repetition does not induce a similar response decrement when variable ISIs are used is a strong indication that the decrement observed when constant stimulation rates are used is indeed at least partially related to the loss of novelty or the higher expectancy of the subsequent stimulus. Three criteria have been established for response decrements to be considered as truly reflecting habituation (Thompson and Spencer 1966). 1. The response decrement related to habituation is expected to follow a negative exponential function of the number of stimulus repetitions. 2. An increase in responding (i.e. ‘response recovery’) should be observed consecutive to the insertion of a change stimulus inserted in a train of repeated stimuli. 3. Following the insertion of a change stimulus, an increase in responding should be observed in response to the previously habituated stimulus (i.e. ‘dishabituation’). The response decrement following stimulus repetition is a well established phenomenon. However, while some studies have found the decrement to be maximal already for the second stimulus, suggesting that the decrement is related to refractoriness and not to habituation, other studies have suggested the opposite, showing the decrement to increase across repetitions and reaching asymptote only at the third or fourth stimulus in the train (see Budd et al. 1998 for a review). 70
Although response recovery of the vertex potential in response to a change stimulus has been established in both the auditory and the somatosensory modality, such a recovery could be explained both by processes of habituation and of refractoriness, either as a result of the incomplete overlap between the neural elements underlying the response to the change stimulus and that underlying the response to the preceding stimulus (Butler and Jones 1968; Naatanen 1988), or to the novelty associated with that change (Sokolov 1975). However, the fact that response recovery may be observed in response to a change stimulus whose only difference with the standard stimulus is, for example, its occurrence at a slightly different time interval (Sable et al. 2003), strongly suggests that response recovery cannot be completely explained by a mismatch between the neural populations elicited by the standard stimulus and that elicited by the change stimulus. Whether dishabituation phenomena may be observed remains equivocal. Indeed, most studies have failed to observe this phenomenon (Fruhstorfer 1971; Woods and Elmasian 1986; Barry et al. 1992; Budd et al. 1998). 6.3
Cognitive and attentional factors
6.3.1 Vigilance Vigilance, largely synonym to arousal, alertness, or sustained attention, would involve processes related to maintaining behavioral goals over time. These processes would also be implicated in the regulation of the sleep-wake cycle. 6.3.1.1
Vertex potentials
Both auditory and somatosensory vertex potentials have been shown to be modulated by the level of vigilance. Indeed, numerous studies have reported that an increase in the general level of attentiveness resulted in an increase in the amplitude of the vertex N1 component. On the contrary, it is well accepted that during the process of falling asleep, the auditory N1 vertex potential declines in amplitude (Ogilvie et al. 1991; Bastuji et al. 1995; Nordby et al. 1996). Furthermore, during non-REM sleep, the auditory N1 component is 71
described as even more attenuated (Nielsen-Bohlman et al. 1991) and may even reach near baseline levels (Paavilainen et al. 1987). This progressive decrease of N1 amplitude parallels a progressive slowing of behavioral response times (Ogilvie et al. 1991). For this reason, the decline in N1 amplitude observed during the process of falling asleep has often been interpreted as resulting from a progressive decline of the subject’s arousal level. Similarly, an attenuation, or even a disappearance, of the auditory vertex potential complex has been described when sedation or drowsiness are pharmacologically induced by benzodiazepines or general anesthesia (Plourde and Picton 1991; Rockstroh et al. 1991; Van Hooff et al. 1995).
Fig. 1-23. Grand average waveforms (CZ-A1A2) of auditory vertex potentials (N1-P2) during wakefulness and stage 2 sleep recorded at electrode CZ. From Crowley and Colrain (2004).
While it is commonly accepted that the amplitude of the auditory N1 is reduced during drowsiness and may even reach baseline levels during nonREM sleep, results concerning the auditory P2 vertex potential are much more equivocal. Indeed, during the process of falling asleep, numerous studies (Nielsen-Bohlman et al. 1991; Ogilvie et al. 1991; Winter et al. 1995; Crowley and Colrain 2004) have shown that the amplitude of the auditory P2 appears, paradoxically, to increase (see figure 1-23). However, it should be noted that this sleep-induced enhancement of P2 amplitude was not reported by all
72
studies (Salisbury and Squires 1993). De Lugt et al. (1996) examined the influence of sleep on the peak-to-peak amplitude of the auditory N1-P2 complex. Results of this analysis revealed no differences across all sleepwake states from relaxed wakefulness to slow-wave sleep. Campbell et al. (1992) suggested that it is neither the N1 nor the P2 component which is affected by sleep but rather that a slow negative wave, overlapping the N1-P2 response, is removed at sleep onset. An opposite interpretation, proposed by Naatanen and Picton (1987), is that during sleep, the stimulus evokes an additional slow positive component. 6.3.1.2
LEPs
It has been repeatedly shown that both late and ultra-late LEPs are strongly attenuated by decreases in arousal (Bromm and Treede 1991; Arendt-Nielsen 1994; Weiss et al. 1997; Bromm and Lorenz 1998). Moreover, due to the longduration and monotony of experimental recordings, declines of vigilance most probably contribute to the often observed progressive amplitude decrement of LEP responses. Beydoun et al. (1993) compared late LEP responses under different states of arousal. LEPs were recorded in subjects after one day of sleep deprivation. Subjects were allowed to fall asleep during the experiment. When subjects were becoming drowsy*, a marked decrease of N2-P2 peak-to-peak amplitude was reported. Furthermore, when subjects were in sleep stage 2†, the laser stimulus did not evoke quantifiable LEPs. Similarly, decreases of LEP amplitude have also been shown to accompany sedation and drowsiness when induced pharmacologically. Indeed, Zaslansky et al. (1996a) showed that intravenous administration of benzodiazepines could induce a marked attenuation of the late laser-evoked P2 component.
*
drowsiness was defined on the basis of a drop-out in EEG alpha-activity and the appearance
of lateral eye movements. †
Sleep stage 2 was defined on the basis of the appearance of sleep spindles on the EEG
73
Responses to selective activation of C-fibers have also been shown to be very sensitive to the level of arousal. Qiu et al. (2002) examined ultra-late LEP responses elicited by selective activation of C-fibers under different states of vigilance (figure 1-24). They showed an almost complete disappearance of the ultra-late LEP during drowsiness and were not able to record any quantifiable C-fiber related potentials during sleep stage II.
Fig. 1-24. Effects of the level of arousal on ultra-late laser-evoked potentials in two subjects (CZ-A1A2). The right hand was stimulated during control (wakefulness), drowsiness, and stage 2 sleep. The ultra-late P2 was strongly attenuated or absent during drowsiness and stage 2 sleep. Waveforms recorded during drowsiness and sleep also appeared smoother due to a reduction in EMG noise (from Qiu et al., 2002).
Therefore, it appears that such as auditory and somatosensory vertex potentials, laser-evoked potentials are sensitive to the level of arousal. Both A -fiber and C-fiber related LEP responses are strongly attenuated during drowsiness and tend to disappear completely during non-REM sleep. However, the observation that the process of falling asleep leads to an apparent increase of the P2 component elicited by auditory stimuli but on the contrary, leads to an almost complete disappearance of the P2 component elicited by laser stimuli argues against the hypothesis that both components could be completely explained by the activation of common generators. Based on the reports that P3a-like electrophysiological components (i.e. the ‘P400 effect’ described by Legrain et al. 2002; 2003a; 2003b; see also section 6.3.4 of Introduction) may significantly contribute to the measured laser-evoked P2 74
amplitude, a possible explanation to this discrepancy could be put forward. Indeed, the P3a component, which is hypothesized to reflect processes related to exogenously-triggered involuntary reorientations of attention, has been shown to be strongly affected by decreases in arousal (Plourde and Picton 1991; Salisbury et al. 1992; Plourde et al. 1993; Bastuji et al. 1995; Winter et al. 1995; Gosselin et al. 2005). Therefore, the attenuation of P2 LEP amplitude related to a decrease in arousal could, at least partially, result from the decrease of an embedded laser-evoked P3a-like component. 6.3.2 Selective attention Selective attention, also referred to as focused attention, would allow biasing or filtering relevant against irrelevant sensory input. This attentional filtering is often considered as a ‘regrettable necessity’ required for limited processing resources to cope with the huge amount of sensory input arising simultaneously from different sensory modalities and locations (Desmedt et al. 1983; Desmedt and Tomberg 1989; Garcia-Larrea et al. 1991; Eimer and Forster 2003). The gating of afferent flow would begin at early, modalityspecific, processing stages. Dayan et al. (2000) proposed that selective attention was not necessarily related to a mean of dealing with resource constraints. Indeed, if one considers that the goal of sensory processing is to get answers to a particular question posed by the environment, selection of inputs appropriate to the question would be beneficial whether or not processing resources are limited. Studies which examine the effects of selective attention on event-related potentials typically compare responses elicited by attended stimuli to that elicited by unattended stimuli. In most of these studies, the effect of selective attention is then assessed by computing unattended – attended difference waveforms. Most studies have focused on the effect of selective attention within the same sensory modality (i.e. intra-modal selective attention; e.g. attending or ignoring the spatial location of the stimulus or a specific attribute of the stimulus). A fewer number of studies have examined the effects of selective attention across different sensory modalities (i.e. inter-modal
75
selective attention; e.g. attending or ignoring stimuli from a specific sensory modality). 6.3.2.1
Selective attention within the auditory modality
The effects of intra-modal selective attention on auditory event-related potentials have been most extensively studied. Most of these studies have described a consistent negative inflection occurring at latencies ranging between 50 – 150 ms after stimulus onset (e.g. Hillyard et al. 1973; Schwent and Hillyard 1975; Hansen and Hillyard 1980; see Figure 1-25). This negative activity was described when subjects attended to the spatial location of the auditory stimulus but also when they attended to a specific acoustic frequency. This negative enhancement was initially interpreted as resulting from an increase of the auditory N1 component elicited by attended stimuli as compared to that elicited by unattended stimuli (Hillyard et al. 1973; Schwent and Hillyard 1975). Indeed, it is generally accepted that selective attention can enhance the receptivity of the cortical networks implicated in the processing of attended inputs (i.e. ‘sensory gain’ hypothesis). However, Naatanen et al. (1978) proposed that this negative enhancement does not reflect an increase of the auditory N1 component per se but instead that it reflects an independent but overlapping electrophysiological component, originating from distinct cortical areas, and referred to as ‘negative difference’ (Nd). The Nd would reflect processes specifically related to selective attention and labeled ‘processing negativities’ (Naatanen et al. 1980; Naatanen and Picton 1987; Naatanen 1990). Processing negativities would involve the comparison of incoming inputs to an attentional trace formed by prior presentations of the attended stimulus (Naatanen et al. 1993). Inputs matching this attentional trace would be further processed while inputs mismatching this template would be fully or partially rejected from higher-order processing.
76
Right ear stimulus
Left ear stimulus
Attend right Attend left
Fig. 1-25. Modulation of late auditory evoked-potentials by selective spatial attention. The amplitude of the auditory N1 component was enhanced when attention was focused on the stimulated ear (thick line) as compared to when it was focused on the other ear (thin line). Adapted from Hillyard et al. (1973).
Whether the effects of selective attention, observed in the 50 – 150 ms range, reflects facilitation of primary ‘sensory’ processes, occurrence of additional, temporally overlapping, purely ‘endogenous’ processes, or a combination of both, has been a source of substantial controversy. Several studies have provided results indicating that the electrophysiological activity related to selective attention could originate from several, overlapping, neural populations (Garcia-Larrea et al. 1995; Kekoni et al. 1996).
77
6.3.2.2
Selective attention within the somatosensory modality
Selective attention has been shown to evoke similar negative enhancements in the somatosensory modality (Desmedt and Robertson 1977; Michie et al. 1987; Garcia-Larrea et al. 1995). Indeed, such as the auditory N1, selective spatial attention effects have been shown to modulate the somatosensory N1 component (Desmedt and Robertson 1977; Josiassen et al. 1982; Desmedt et al. 1983; Michie et al. 1987; Desmedt and Tomberg 1989; Papanicolaou et al. 1989; Garcia-Larrea et al. 1991; Ito et al. 1992; Garcia-Larrea et al. 1995). The effect of spatial attention on the somatosensory N1 was initially demonstrated by Desmedt and Robertson (1977). This study showed that the N1 components elicited by stimuli presented to an attended finger were enhanced as compared to those elicited by stimuli presented to an unattended finger. It has been suggested by Garcia-Larrea et al. (1995) that at least part of the enhancement of the somatosensory N1 observed when stimuli are presented at an attended location could be related to a ‘processing negativity’ similar to that described by Naatanen (1980) in the auditory modality. In addition to affecting the later N1 vertex component, selective spatial attention has been shown to modulate earlier somatosensory P27 and P45 components (Josiassen et al. 1982; Desmedt et al. 1983; Desmedt et al. 1984). This earlier effect of spatial attention was clearly demonstrated by Garcia-Larrea et al. (1991) who showed that non-target somatosensory stimuli delivered at an attended location could elicit enhanced P27 and P45 components. Therefore, it appears that intra-modal selective attention within the somatosensory modality could induce an enhancement of somatosensory evoked potentials through a ‘sensory gain’ mechanism but also elicit additional purely endogenous components, similar to the ‘processing negativities’ described in the auditory modality. 6.3.2.3
Inter-modal selective attention
Most studies have examined the effects of selective attention within the same sensory modality (i.e. intra-modal selective attention). However, a number of studies (Woods et al. 1992; Alho et al. 1994; de Ruiter et al. 1998; Talsma 78
and Kok 2001; Hotting et al. 2003) have also examined the effects of intermodal selective attention (i.e. attending or ignoring stimuli from a specific sensory modality). Results of these studies have revealed that attending to a specific sensory modality could induce negative enhancements of eventrelated responses similar to that described as related to intra-modal selective attention. Whether the mechanisms involved in inter-modal selective attention are shared with that involved in intra-modal selective attention remains a matter of debate. 6.3.2.4
Selective attention within the nociceptive modality
Numerous studies have compared LEPs with attention directed either towards or away from the laser stimulus (Beydoun et al. 1993; Siedenberg and Treede 1996; Zaslansky et al. 1996b; Garcia-Larrea et al. 1997; Yamasaki et al. 1999; Friederich et al. 2001). All these studies have consistently reported that attending to the laser stimulus could induce a strong enhancement of the vertex N2-P2 complex. Results of these studies have also suggested that the earlier N1 LEP component was mostly unaffected by the focus of attention (Garcia-Larrea et al. 1997). However, determining the exact causes underlying these LEP amplitude modulations is rendered difficult by the fact that, in most experimental paradigms, several attentional factors were concurrently manipulated (see Table 1-1): -
In most of these studies (Beydoun et al. 1993; Plaghki et al. 1994; Siedenberg and Treede 1996; Zaslansky et al. 1996b; Garcia-Larrea et al. 1997; Yamasaki et al. 1999; Friederich et al. 2001; Nakamura et al. 2002; Schlereth et al. 2003), LEPs were compared across different experimental conditions presented within different recording blocks. As tasks requested within each experimental conditions differed (e.g. active counting of incoming stimuli vs. passively waiting for the recording sequence to end), use of such paradigms could have led to significant variations in the level of arousal. Such changes in arousal could therefore have contributed to the observed LEP differences.
-
In a number of studies (Beydoun et al. 1993; Towell and Boyd 1993; Plaghki et al. 1994; Kanda et al. 1996; Siedenberg and Treede 1996; Zaslansky et al. 1996b; Garcia-Larrea et al. 1997; Yamasaki et al. 79
1999; Legrain et al. 2002; Nakamura et al. 2002; Legrain et al. 2003a; Legrain et al. 2003b; Schlereth et al. 2003), subjects were required to detect and react to the attended stimulus. Observed LEP differences may, in these conditions, have been related to the task-relevance or target nature of the attended stimulus. -
Furthermore, in some studies (Beydoun et al. 1993; Towell and Boyd 1993; Plaghki et al. 1994; Siedenberg and Treede 1996; Zaslansky et al. 1996b; Garcia-Larrea et al. 1997; Yamasaki et al. 1999; Friederich et al. 2001; Nakamura et al. 2002; Schlereth et al. 2003), attention was shifted across both a different sensory modality and a different spatial location. Under these conditions, both inter-modal and intra-modal selective attention effects could have modulated LEP responses.
Fig. 1-26. Grand-average of laser-evoked potentials recorded from left and right hand stimulation at electrode CZ. Frequent and non-frequent stimulus intensities were applied to the left and right hands. Subjects selectively attended either the left or right hand (subjects were requested to count rare targets occurring at the attended hand). Attending to the hand led to an enhancement of LEP N1 and N2 components, regardless of the fact the stimulus was a target and regardless of probability of occurrence. Rare and attended stimuli elicited an additional P3b-like component (P600). Adapted from Legrain et al. (2002).
80
EEG
Hand immersed in cold water Perform a reaction-time and intensity evaluation task.
Heterotopic Nociceptive Conditioning (HNCS)
Perform a mental arithmetic task. Perform a reaction-time and intensity evaluation task.
Mental arithmetic Task (MT)
?
Arousal
?
?
EEG (upper lip stimulation)
presented at other side of left hand
Infrequent target (P=.15)
presented at one side of left hand
Before and after ‘conditioning’ with either MT or HNCS. Perform a reaction-time and intensity evaluation task.
Control
EEG
Frequent non-target (P=.85)
Plaghki et al. (1994)
Oddball
Passive stimulation of dorsum of left hand.
Control
Towell and Boyd (1993)
Appearance of sleep spindles.
Sleep stage 2
One day of sleep deprivation, subjects allowed falling asleep.
Drowsiness
Perform a mathematical operation.
Distraction
Evaluate intensity of perceived stimuli
Attention
Beydoen et al. (1993)
Table 1-1. Attend modality
Attend spatial
?
?
Stimulus Relevance Novelty
N2-P2
(-42%)
(-38%)
N2-P2
N2-P2 =
P2 + P600
no N2-P2
N2-P2
LEP amplitude
EEG
Presented at other side of left hand
Infrequent target (P=.20)
Presented at one side of left hand
EEG
Oddball
Oddball
Infrequent non-target
Count frequent targets presented at one location of the hand
Frequent target
Count infrequent targets presented at one location of the hand
Infrequent target
Frequent non-target
Perform arithmetical operation during presentation of stimulus.
Distraction
Perform arithmetical operation after perception of stimulus.
Attention
Zaslansky et al. (1996b)
IV Benzodiazepine administration. Evaluate average intensity of perception after each block.
Sedation
IV placebo administration. Evaluate average intensity of perception after each block.
Placebo
Evaluate average intensity of perception after each block.
Control
EEG
Frequent non-target (P=.80)
Zaslansky et al. (1996a)
(with and without a motor task)
Oddball
Kanda et al. (1996)
?
?
P2
P2
P2
P2
P2
N2-P2 = + P600
EEG
MEG / EEG
Memorize and subsequently report a series of numbers.
Distraction – memorization
Perform arithmetical operations during presentation of stimulus.
Distraction – calculation
No task
Control
Yamasaki et al. (1999)
Occasional gaps in concurrently presented white noise Attend and count auditory gaps. Report mean intensity of laser stimuli at end of each run. * Decrease of subjective pain rating.
Distraction
Occasional gaps in concurrently presented white noise Attend and count laser stimuli. Report mean intensity of laser stimuli at end of each run.
Attention
EEG
Count / press a button at infrequent targets at one location (foot)
Infrequent target
Frequent non-target
Garcia-Larrea et al. (1997)
(with and without a motor task)
Oddball
Read a book.
Distraction
Neutral
Siedenberg and Treede (1996)
=
?
?
?
=
=
=
N2-P2 1M =
N2-P2 1M =
N2-P2 N1 =
P2 + P600
P2
P2
constant 3s ISI
Attend at one side and count rare intensity deviant targets (intensity discrimination)
Oddball
EEG = =
=
ATTENDED – FREQUENT
ATTENDED – INFREQUENT (target)
=
UNATTENDED – INFREQUENT
UNATTENDED – FREQUENT
Legrain et al. (2002)
Match intensity of laser stimuli with frequency of preceding auditory stimulus.
“High-level” of attention ?
?
?
=
=
=
=
?
?
?
1M
1M
? (n.s.) P2
N1 N2 P2 + P600
N1 N2
P2
(same as mid-level)
N2
N2 = P2 =
The experiment did not control for general arousal and spatial attention between “low-level” and “mid-level” or “high-level” of attention. There is no evidence indicating a difference in attention between “mid-level” and “high-level” of attention.
*
EEG
EEG
Rate intensity of laser stimulus preceded by an auditory stimulus.
“Mid-level” of attention
No task.
“Low-level” of attention
*
Nakamura et al. (2002)
Listen to a story with subsequent memorization test.
Distraction (DA)
Suggestion of hypnosis analgesia.
Suggestion of hypnosis analgesia (HA)
No task
Control
Friederich et al. (2001)
(strong and weak stimulus intensity)
ATTENDED – INFREQUENT (TARGET)
(strong and weak stimulus intensity)
ATTENDED – FREQUENT
(strong and weak stimulus intensity)
UNATTENDED – INFREQUENT
=
=
=
=
=
=
=
= (P3a?)
(P3a?) (both strong and weak)
P600
(only for strong)
P2
P400 (fronto-central P3a)
(only for strong)
P2
P600 Parietal P3b
Table 1-1 compares the different experimental conditions of several studies which examined the modulation of LEPs by attention. For each condition, it was assessed whether one of the following attentional factors had been manipulated. (1) Arousal: the general level of arousal. (2) Attend modality: several studies compared LEPs when subjects were either attending to the ‘nociceptive’ modality or attending to another sensory modality. In these studies, LEP modulations may have been related to effects of inter-modal selective attention. (3) Attend spatial: several studies compared LEPs when subjects were either attending to the spatial location of the stimulus or attending to another body part. In these studies, LEP modulations may have been related to effects of spatial attention. (4) Stimulus relevance: in a number of studies, stimulus relevance was manipulated by the task. (5) Novelty: by contrasting the probability of occurrence of stimuli, several studies examined the effects of novelty or unexpecteness. The last column outlines reported LEP modulations. N1, N2, and P2 refer to the amplitude of LEP components. 1M refers to the amplitude of the earliest (~170 ms) MEG component. P600 and P400 refer to two additional positive components which sometimes followed the P2 (see sections 6.3.3 and 6.3.4 of Introduction).
constant 3s ISI
Attend at one side and count rare intensity deviant targets (intensity discrimination)
Oddball
(strong and weak stimulus intensity)
UNATTENDED – FREQUENT
EEG
INFREQUENT
UNATTENDED – (distractor: strong)
constant 3s ISI
Legrain et al. (2003b)
INFREQUENT
ATTENDED – (target: strong)
Attend at one side and count rare (strong) targets
Oddball
EEG
ATTENDED – FREQUENT (nontarget: weak)
50 stimuli of constant intensity, no task assigned
Control
Legrain et al. (2003a)
In a recent set of studies, Legrain et al. (2002; 2003a; 2003b) specifically examined the effect of the spatial direction of attention within the nociceptive modality. These studies showed that all LEP negativities, (i.e. the N2 but also the N1 component) were increased in response to laser stimuli at selectively attended body locations (see figure 1-26). This modulation was present whether or not attended stimuli were targets (i.e. relevant to the task). On the contrary, the laser-evoked P2 component was unaffected by the spatial location of the attentional focus. Although the topographies of the enhanced N1 and N2 LEP components were unchanged by selective spatial attention, suggesting a true enhancement of these components, some investigators have interpreted the concurrent enhancement of both LEP negativities as resulting from an overlapping ‘processing negativity’, analogous to the one reported in other sensory modalities (Lorenz and Garcia-Larrea 2003). In addition to being sensitive to the locus of spatial attention, the magnitude of LEP components may be significantly modulated by a modality-specific ‘sensory gain’ effect. Indeed, although the possible contributions of factors such as task-relevance and arousal limit the interpretability of results, several studies have shown that laser stimuli yield enhanced responses when subjects attend to the nociceptive modality as compared to when they attend to another sensory modality (Beydoun et al. 1993; Towell and Boyd 1993; Plaghki et al. 1994; Siedenberg and Treede 1996; Zaslansky et al. 1996b; Garcia-Larrea et al. 1997; Yamasaki et al. 1999; Friederich et al. 2001; Nakamura et al. 2002; Schlereth et al. 2003). The modulations of LEPs by selective attention therefore appear very similar to that of somatosensory and auditory vertex potentials. Effects could consist in a combination of (1) an enhancement of the late negativities, including the vertex negativity, possibly related to an increased receptivity of the underlying cortical structures (sensory gain hypothesis) and (2) the occurrence or enhancement of an additional ‘processing negativity’, reflecting purely endogenous processes related to the comparison of incoming sensory input to the temporarily memorized characteristics of the attended input.
6.3.3 Task relevance Sutton et al. (1965) first described that when manipulating task relevance and occurrence probability of auditory stimuli, auditory events could elicit an additional positive component. Desmedt et al. (1965) almost simultaneously confirmed this observation in the somatosensory modality. This response, often referred to as the P3b component, occurs approximately 300 – 350 ms after the onset of an auditory stimulus and 350 – 450 ms after the onset of a somatosensory or visual stimulus (Johnson 1986; Picton 1992). To elicit a P3b component, the evoking stimulus must be infrequent and subjects must be actively involved in its detection. It is currently accepted that P3b waves represent late stages of information processing. To explain their functional significance, two leading hypotheses have been put forward. The first proposes that the P3b reflects the updating of working memory following the arrival of new information or ‘context updating’ (Donchin and Coles 1988). The second proposes that the P3b reflects the closure of the processing of information or ‘context closure’, occurring when expectations are terminated (Desmedt 1980; Verleger 1988). 6.3.3.1
The laser evoked P600
A number of studies have used oddball paradigms* to search for the presence of laser-evoked P3b-like responses (Towell and Boyd 1993; Kanda et al. 1996; Siedenberg and Treede 1996; Zaslansky et al. 1996a; Legrain et al. 2002; Legrain et al. 2003a). In most of these studies, frequent and infrequent stimuli differed by their spatial location (Towell and Boyd 1993; Kanda et al. 1996; Siedenberg and Treede 1996; Zaslansky et al. 1996b). To allow dissociating between the effects of spatial attention and that of task relevance, some more recent studies have used different stimulus intensities to define stimulus deviance (Legrain et al. 2002; Legrain et al. 2003a; Legrain et al. 2003b).
*
The oddball paradigm is the most used paradigm to evoke P300 responses. In this
paradigm, two physically different stimuli are sequentially presented with contrasting probabilities. To focus the subjects attention towards the infrequent stimulus, subjects are usually requested to perform a task related to the detection of the infrequent ‘target’ stimulus.
87
Fig. 1-27. Latency-corrected grand average waveforms (eight subjects) of ultra-late laser-evoked potentials recorded using a very small stimulus surface area. Stimuli were presented within an oddball paradigm at two different fingers of the left hand. Subjects were requested to press a button upon perceiving the infrequent target stimulus. A1A2 was used as reference. Individual averages were adjusted on the positivity of the N2-P2 complex. Left and right panels show non-target and target stimulus conditions respectively. Note the larger amplitude of the ultra-late P2 and the appearance of an additional positive component (P3), following the N2-P2 complex in the target condition. This additional positivity was maximal at electrode PZ. Vertical dashed lines show the time-points of plotted scalp maps. From Opsommer et al. (2003).
88
Task relevance was obtained by asking subjects to silently count or press a button when perceiving the infrequent target stimulus. Most investigators* have reported that infrequent task-relevant laser stimuli indeed elicited an additional positive component (see figure 1-26) whose topography was similar to that of the P3b component elicited by other sensory modalities (Towell and Boyd 1993; Kanda et al. 1996; Siedenberg and Treede 1996; Legrain et al. 2002; Legrain et al. 2003a). This component, occurring approximately 600 ms after stimulation of the hand, could clearly be distinguished from the earlier P2 component. Using a similar oddball paradigm, Opsommer et al. (2003) showed that selective activation of C-fiber nociceptors could also elicit a P3blike response, occurring approximately 1300 ms after stimulation of the hand (figure 1-27). 6.3.4 Stimulus-driven attentional capture It has often been hypothesized that the enhancement of vertex potentials elicited by the first stimulus in a train of repeated stimuli was related to the triggering of an initial orienting-response (Kenemans et al. 1989). Similarly, several investigators have proposed that the ‘response recovery’ of vertex potentials observed when a change stimulus is inserted within that train of stimuli could be related to processes triggered by the novelty or deviance associated with that change. In the auditory modality, numerous studies have shown that physically deviant sounds presented within a repetitive sequence (e.g. sounds differing in pitch, *
Zaslansky et al. (1996) proposed that the laser-evoked P2 component could reflect
processes functionally equivalent to that of the P3b elicited by other sensory modalities. Their hypothesis was based on the finding that infrequent target laser stimuli elicited a larger P2 component than frequent or ignored laser stimuli. However, this view is contradicted by two arguments. The first argument is based on the fact that in this study, both target and nontarget infrequent stimuli elicited enhanced P2 components. Indeed, if the P2 did correspond to a P3b-like component, only infrequent target stimuli should have produced an enhancement of its amplitude. The second argument is based on the topography of the laser-evoked P2 component. Indeed, the topography of P3b components elicited by other sensory modalities has been shown to be more posterior than the vertex-centered topography of the laserevoked P2.
89
intensity, duration, location, or timing) could elicit a negative inflection of the EEG, referred to as a ‘mismatch negativity’ (reviewed in Naatanen et al. 1992). The mismatch negativity or MMN is elicited even when the subject’s attention is diverted from the sound (figure 1-28). For this reason, it has been suggested that the MMN reflects an automatic form of sensory analysis. To explain this component, Naatanen proposed that for the purpose of detecting changes in the auditory milieu, the brain automatically forms a short-term memory trace of auditory features which is then continuously compared to the incoming stream of sensory information. Detection of such changes would trigger processes reflected by the MMN component.
Location-MMN Frequency-MMN
Left (L1)
Mid (FZ)
Right (R1)
Fig. 1-28. Mismatch negativities (MMN) responses obtained by subtracting the average waveform to standard tones from the average waveform to deviant tones. Filled line: deviance consisted in the occasional occurrence of a spatially displaced auditory tone. Dotted line: deviance consisted in the occasional occurrence of a differently pitched tone. Adapted from Schröger and Wolff (1997).
Refractoriness of brain processes could be an alternative hypothesis explaining why the deviant stimulus elicits a greater response than the standard stimulus. Indeed, the neural elements activated by the deviant stimulus could be expected to differ slightly from that activated by the preceding standard stimulus. In other words, the MMN elicited by deviant stimuli could be related to the activation of afferent elements which have not been put in a ‘refractory state’ by the preceding stimulus. However, there is strong evidence indicating that the MMN component cannot be attributed to ‘refractoriness’. Indeed, it was shown that the occasional omission of a second tone of two closely-paced tones elicits a MMN response (Yabe et al.
90
1997). Furthermore, it was also shown that a tone repetition in a sequence of steadily descending tones elicits a MMN (Tervaniemi et al. 1994). Several studies have suggested that the greater part of the auditory MMN is generated in the auditory cortex (Alho 1995). However, some investigators have proposed that bilateral frontal generators could also contribute to the MMN component (Giard et al. 1990). This frontal source was hypothesized to play a role in the initiation of an involuntary attention switch triggered by a sound change preperceptually detected in auditory cortices. In support of this hypothesis, it was shown that the signals generated by these frontal generators appear with a slight time-delay as compared to those generated by the bilateral temporal generators (Rinne et al. 2000). Whether deviant somatosensory stimuli may elicit EEG changes similar to the MMN component elicited by deviant auditory stimuli is still not clearly determined. Nevertheless, it should be noted that in a study examining eventrelated potentials elicited by deviant and ignored vibratory stimuli, Kekoni et al. (1997) proposed that the earlier part of the vertex negativity (N120) could reflect a MMN-like activity. This somatosensory MMN component was hypothesized to originate from somatosensory-specific cortical regions. In addition, numerous studies (Courchesne et al. 1975; Squires et al. 1975; Yamaguchi and Knight 1991; Escera et al. 1998; Katayama and Polich 1998) have shown that deviant or intrusive auditory, visual, or somatosensory stimuli which occur outside the focus of attention may evoke an additional positive component. This component, often referred to as ‘P3a’, has an earlier latency and a more frontal scalp distribution than the P3b component elicited by taskrelevant and infrequent stimuli. The P3a component is hypothesized to index an involuntary attentional-orienting reaction triggered by the detection of a sudden change in the environment. 6.3.4.1.1 The laser-evoked “P400 effect” In a recent study, Legrain et al. (2002) showed that rare intensity deviant laser stimuli presented within a stream of standard stimuli could elicit an additional 91
Fig. 1-29. Grand-average of laser-evoked potentials recorded at three midline electrodes (FZ, CZ, PZ vs. A1A2). Strong and weak stimuli (either frequent or non-frequent) were applied to the left and right hands. Subjects were requested to count the nonfrequent stimuli occurring at the attended hand (target stimuli). The main positive peak, was larger in amplitude for non-frequent strong than for frequent strong stimuli (P400). It was not larger for non-frequent weak than for frequent weak stimuli. An additional parietal positivity was evoked at a later latency by both strong and weak target stimuli (P600).
positive deflection, occurring approximately 400 ms after stimulus onset (see figure 1-29). As this activity was elicited by deviant stimuli presented both within and outside the spatial focus of attention, it was hypothesized that it could reflect processes related to the P3a component described in other sensory modalities. In other words, the processes underlying this “P400 effect” would be involved in an involuntary orientation of attention. In a later study comparing responses elicited by strong and weak deviant laser stimuli, Legrain et al. (2003b) showed that only laser stimuli of strong intensity could elicit this “P400 effect”. This was interpreted as an indication that weak stimuli
92
were not salient enough to induce involuntary attentional switching. This P3alike component strongly overlapped with the later part of the LEP P2 component. Therefore, it is most probable that when large and unpredictable inter-stimulus intervals are used, such a “P400 effect” significantly contributes to the P2 component. Such a contribution could explain the results of the study by Zaslansky et al. (1996b), which showed that infrequent target laser stimuli elicited a larger P2 component than frequent and ignored laser stimuli. As discussed in sections 5.3.2 and 6.3.1 of Introduction, such a contribution could also account for some of the differences between the laser-evoked P2 and the somatosensory or auditory evoked P2 components. 6.4
In brief
Late and ultra-late LEP components appear to be equally modulated by experimental parameters such as stimulus intensity, general level of arousal, and task relevance. These observations provide further evidence suggesting that both LEP responses reflect the activation of common neural generators. Furthermore, experimental manipulations have a similar influence on the vertex potentials which may be recorded in other sensory modalities. Indeed, such as LEPs, vertex potentials are highly correlated with the intensity of the evoking stimulus, are highly sensitive to stimulus repetition, are sensitive to the general level of arousal, are modulated by selective attention, and are hypothesized to reflect, at least partially, processes related to the triggering of orienting responses related to the novelty or deviance of the evoking stimulus. Therefore, it is most probable that at least a fraction of the neural generators underlying late and ultra-late LEPs share common sources with the late vertex potentials recorded within other sensory modalities.
93
7 Why C-nociceptive input elicits an ultra-late LEP only when concomitant activation of A -fibers is avoided To explain why C-nociceptive input appears to elicit an ultra-late LEP only when the concomitant activation of A -nociceptors is avoided, three hypotheses have often been put forward. These different hypotheses have important consequences regarding the functional significance ascribed to the processes underlying both late and ultra-late LEPs. 7.1
Non-stationarity of the C-fiber afferent volley
At first, the absence of obligatory ultra-late LEP components was assumed to result from the non-stationarity of the C-fiber afferent volley. Indeed, microneurographic recordings have shown that the conduction velocity of Cfibers can vary greatly, and moreover, is reduced by repetitive activation (Valbo et al. 1979). Therefore, several investigators had proposed that the poorly synchronous nature of C-fiber afferents could explain why their activation did not elicit consistent ultra-late LEP brain responses (Bromm and Treede 1987b; Arendt-Nielsen 1990). Indeed, it was likely that an important latency-jitter of C-fiber input could prevent conventional time-averaging techniques from revealing C-fiber related brain potentials (see Appendix A). To reduce the negative influence of temporal jitter, several investigators (Carmon et al. 1980; Bromm and Treede 1987b; Arendt-Nielsen 1990; Purves and Boyd 1993) have used latency-correction algorithms (e.g. Woody filtering, time-shifted averaging). Although these techniques may generate artifactual potentials, when applied to the analysis of stimuli concurrently activating A and C-fiber nociceptors, they repeatedly failed to detect consistent stimulusrelated signal changes within the time-window compatible with the conduction velocity of C-fibers. Furthermore, unless one would assume that the latencyjitter of C-fiber afferents is transiently increased by a preceding A -fiber volley, such a latency-jitter could not explain why ultra-late LEP responses are consistently recorded if C-nociceptors are activated in isolation. In addition to evoked potentials, stimuli may induce transient enhancements or attenuations of ongoing EEG oscillations. These changes can be revealed 94
by methods based on the time-frequency decomposition of EEG epochs (see Appendix A). Using related techniques, several studies have shown that laser stimuli may induce prolonged modulations of the EEG frequency spectrum (Arendt-Nielsen 1990; Ploner et al. 2002). As these long-lasting cortical responses are spread across the time-window compatible with the conduction time of C-fibers, it was proposed that they may reflect activity related to the processing of C-fiber input. However, it remains unclear as to whether these EEG changes, which were observed in response to stimuli concomitantly activating A - and C-nociceptors were indeed elicited by the later-arriving Cfiber volley or whether they reflected long-lasting activity elicited by the firstarriving A -fiber volley. 7.2
A -fiber mediated spinal inhibition of C-fiber afferent transmission
Several investigators have proposed that A -fiber induced inhibition of C-fiber spinal transmission could explain why C-fiber activation does not elicit ultralate brain responses when A -fibers are concomitantly activated (ArendtNielsen 1990). In other words, the C-fiber afferent volley would be blocked at the spinal level by processes triggered by the shortly preceding A -fiber afferent volley. This hypothesis was based on studies having shown that repetitive stimulation of peripheral nerves in primates could produce a sustained central inhibition of spinal transmission (Chung et al. 1984). Furthermore, these studies had shown that the greatest inhibition of C-fiber transmission was produced by A -fiber stimulation. Accordingly, several behavioral studies in humans, using long-duration tonic heat stimuli, have shown that blocking the peripheral transmission of A -fibers not only completely suppressed the first pain sensation but also consistently increased the second pain sensation (Landau and Bishop 1953; Sinclair and Stokes 1964; Price et al. 1977). However, if the C-fiber afferent volley is blocked at the spinal level, how is it that the laser stimulus clearly leads to the perception of both first and second pain? Consequently, one may question the ability of the very transient A -fiber afferent volleys produced by brief laser heat stimuli to produce a similar inhibition of C-fiber spinal transmission. In fact, psychophysical studies 95
(Chakour et al. 1996; Nahra and Plaghki 2003a) using brief laser stimuli have failed to replicate the enhancement of second pain after A -fiber blockade which had been described by studies using long-duration heat stimuli (see also figure 1-4). Indeed, these studies showed that blocking A -fibers led to an overall decrease of pain sensation related to the disappearance of first pain, but did not show any increase of the C-fiber mediated sensation of second pain. 7.3
Refractoriness of LEP cortical generators
As overviewed in section 5 of Introduction, the similarities between A - and Cfiber event-related responses have led a number of investigators to propose that late and ultra-late LEPs reflect the activation of common generators. To explain why activation of C-fibers appears to evoke an ultra-late LEP response only when the C-fiber afferent volley is not preceded by an A -fiber afferent volley, it has thus often been hypothesized that LEP generators could be subject to ‘refractoriness’ (Bromm and Treede 1987b; Treede et al. 1988b; Magerl et al. 1999; Ploner et al. 2002; Kakigi et al. 2003). Consequently, when A -fibers would trigger a late LEP, the transient ‘refractory period’ of the underlying LEP generators would explain why the later-arriving C-fiber afferent volley does not elicit an ultra-late LEP. The fact that several studies reported that repetition of the laser stimulus induced a significant reduction of LEP amplitude (see section 6.2 of Introduction) could be considered as giving support to this hypothesis. However, it should be emphasized that these studies did not allow distinguishing between a response decrement induced by habituation and related to the loss of novelty or unexpectedness and a response decrement secondary to true refractoriness of underlying neural generators.
96
Chapter 2. Study Objectives
In addition to evoked-potentials, sensory, motor, and cognitive events induce transient enhancements or attenuations of the ongoing EEG oscillatory activity (reviewed in Appendix 1). Hypotheses are that these modulations reflect mechanisms involved in cortical activation, inhibition, and probably binding. As these ongoing EEG oscillations are cancelled-out by conventional timeaveraging procedures, revealing these activities requires the use of alternative techniques based, for example, on the analysis of the EEG power spectrum as a function of time. Using related methods, studies have shown that the laser stimulus induced spectral modulations of the EEG which spread across the time-window compatible with the conduction-time of C-fibers. For this reason, it was proposed that these cortical activities could be related to the central processing of C-fiber input (Arendt-Nielsen 1990; Ploner et al. 2002). However, it remained unclear as to whether these long-lasting EEG changes were indeed related to the later-arriving C-fiber volley or whether they reflected prolonged activity related to the first-arriving A -fiber volley. Furthermore, due to the limited time resolution of the methods used, very transient electrophysiological responses could have been overlooked. The aim of the first study was therefore to develop and apply novel time-frequency signal-processing methods to the analysis of laser-induced EEG changes. Indeed, these more recent methods, based on the wavelet decomposition of the signal, are best suited for examining transient oscillatory changes which, such as the EEG, are widespread in the frequency domain. Results of this study are presented in Chapter 3. A description of the software which was developed for performing these analyses can be found in Appendix C. The most commonly accepted hypothesis to explain why a preceding A -fiber afferent volley apparently leads to the occlusion of the C-fiber ultra-late response relies on the assumption that A - and C-fiber responses are related to the activation of shared cortical generators. Indeed, a number of investigators have proposed that prior activation of these generators by the preceding A -fiber afferent volley could place them in a transient ‘refractory 97
state’ which would explain why the later arriving C-fiber volley is unable to trigger an ultra-late LEP response (Bromm and Treede 1987b; Treede and Bromm 1988b; Magerl et al. 1999; Plaghki and Mouraux 2002; Ploner et al. 2002; Kakigi et al. 2003). Although the ‘refractoriness’ hypothesis agrees with current experimental findings, it has never been directly investigated. The aim of the second study, presented in Chapter 4, was therefore to test this hypothesis using two consecutive but unexpected laser pulses applied to the skin with variable inter-stimulus intervals. In most laser stimulation paradigms, subjects are asked to pay close attention to the upcoming stimulus. Therefore, in most cases where specific methods are used to selectively activate C-nociceptors, attention is selectively focused on the isolated sensation of second pain. However, when the laser stimulus co-activates A - and C-fiber nociceptors, attention may be mostly focused towards the more salient event of first pain, and this even though subjects report perceiving both first and second pain. It was hypothesized that a necessary condition for C-fiber inputs to elicit ultra-late LEP responses could be that attention be entirely focused on that specific sensory channel. In accordance with this hypothesis, Bromm and Treede (1985) reported that if a subject is trained to focus his attention onto second pain, a laser stimulus coactivating A - and C-fiber nociceptors could elicit both late and ultra-late LEP responses. However, the fact that this study relied on results of a single subject and used latency-correction algorithms limits its significance. Therefore, in an attempt to replicate these findings, the third study compared LEP responses to stimuli concomitantly activating A - and C-fiber nociceptors when selective attention was focused towards first pain with LEP responses to the same stimuli when selective attention was focused towards second pain. Results of this study are presented in Chapter 5.
98
Chapter 3. Non-phase locked EEG responses to CO2 laser skin stimulations may reflect central interactions between A and C-fiber afferent volleys
Mouraux A. Guérit J.M. Plaghki L. Non-phase locked electroencephalogram (EEG) responses to CO2 laser skin stimulations may reflect central interactions between A[delta]- and C-fibre afferent volleys. Clin Neurophysiol 2003; 114: 710-722. In addition to evoked-potentials, sensory, motor, and cognitive events may induce transient enhancements or attenuations of ongoing EEG oscillations (referred to as event-related ‘synchronization’ or ‘desynchronization’). Hypotheses are that these modulations reflect mechanisms involved in cortical activation, inhibition, and probably binding (see Appendix A). As ongoing EEG oscillations are not ‘phase-locked’ to the onset of the stimulus, these activities cannot be revealed by conventional time-domain averaging procedures. Therefore, revealing these activities requires the use of alternative methods which are based on the time-frequency decomposition of EEG epochs. Using similar methods, a number of studies have shown that the laser stimulus induced such modulations of the EEG frequency spectrum (Arendt-Nielsen 1990; Ploner et al. 2002). As these long-lasting cortical responses spread across the time-window compatible with the conductiontime of C-fibers, it was proposed that they might reflect central processing of C-fiber input. However, partly due to the limited time-resolution of the methods used in these studies, it remained unclear as to whether these long-lasting EEG changes, produced by stimuli concomitantly activating A - and C-fibers, were indeed related to the later-arriving C-fiber volley or whether they reflected prolonged activity related to the first-arriving A -fiber volley. Furthermore, this methodological limitation could have overlooked very transient electrophysiological responses. The aim of this first study was therefore to develop and apply recent methods based on the prior wavelet decomposition of EEG epochs. Indeed, these more recent methods allow optimizing the time and frequency resolution of the analysis and are therefore best suited for examining transient oscillatory changes in signals which are widespread in the frequency domain. 99
Abstract Objective: By co-activating A - and C-fiber nociceptors, intense CO2 laser heat stimuli produce a dual sensation, composed of first and second pain, but induce only a single A -fiber related late laser evoked potential (LEP). However, when avoiding concomitant activation of A -fibers, C-fiber related ultra-late LEPs are recorded. This poorly understood phenomenon was reinvestigated using a method which, unlike time-domain averaging, reveals EEG changes whether or not phase-locked to stimulus onset. Method: CO2 laser stimuli were applied to the dorsum of the hand. Reactiontime was used to discriminate between A - and C-fiber mediated detections. Analyses were performed using a method based on the time-frequency wavelet transform of EEG epochs. Results: This study revealed (1) a novel non phase-locked component related to the activation of A -fibers occurring at similar latencies as the late LEP and (2) a widespread post-stimulus event-related desynchronization (ERD) induced by both A - and C-fibers. Conclusion: A - and C-fiber related LEPs could be electrophysiological correlates of similar brain processes, which, when already engaged by A fibers, cannot or do not need to be reactivated by the later arriving C-fiber afferent volley. A -fiber related ERD could reflect a transient change of state of brain structures generating these responses.
100
1 Introduction Powerful and brief CO2 laser heat stimuli produce a characteristic dual perception composed of first pain (described as a ‘well localized’, ‘sharp’, ‘pricking’ sensation) and second pain (described as a ‘diffuse’ and ‘burning’ after-sensation). This double pain response was first described by Lewis and Ponchin (1937). Numerous studies have demonstrated that these sensations are related to the selective and concomitant activation of A - and C-fiber nociceptors (Bromm and Treede 1984; Willis 1985; Price 1988). The difference in conduction velocity of A - (~10 m/s) and C- (~1.0 m/s) fibers explains why first and second pain occur at different latencies. The activation threshold of C-fiber nociceptors is lower than that of A -fibers (Treede et al. 1995). Therefore, while high intensity laser stimuli induce the perception of both A -fiber related first pain and C-fiber related second pain, low intensity laser stimuli only induce the perception of second pain. As a consequence, when using a range of stimulus intensities, reaction-time (RT) frequency distribution is bimodal (Campbell and LaMotte 1983; Plaghki et al. 1994; Towell et al. 1996). Laser-evoked brain potentials (LEPs) display components in a time window (~160-390 ms; Bromm and Treede 1984) compatible with the conduction velocity of A -fibers. Several studies demonstrated that this late LEP is a correlate of first pain (for a review see Chen et al. 1998). However, no significant electrical brain activity can be detected at latencies compatible with the conduction velocity of C-fibers (Treede and Bromm 1988b; Towell et al. 1996). Several methods (Bromm et al. 1983; Bragard et al. 1996; Magerl et al. 1999) and pathological conditions (Kakigi et al. 1991; Lankers et al. 1991) allow selective activation of C-fiber nociceptors (for a review see Plaghki and Mouraux 2002). This results not only in the disappearance of first pain and its electrophysiological correlate, the late LEP, but also in the appearance of an ultra-late LEP (~750-1150 ms) compatible with conduction velocities of Cfibers. Why concomitant activation of A - and C-fibers does not allow the individualization of both late and ultra-late LEPs is not well understood. The present study was undertaken to explore a possible interaction occurring 101
between electrical brain activities related to the activation of A - and C-fiber nociceptors. Event-related potentials (ERPs), revealed by time-averaging, result from evoked EEG changes in electric potential that are both ‘phase’* and timelocked to the event onset. Sensory, motor, and cognitive processes can also induce event-related desynchronization (ERD) and synchronization (ERS) corresponding to induced modulations of the amplitude of ongoing EEG oscillations (for a review see Lopes da Silva and Pfurtscheller 1999). Hypotheses are that these modulations reflect mechanisms involved in cortical activation, inhibition, and, probably, binding. Since these ongoing EEG oscillations are not phase-locked to the event onset, they are lost by time averaging. The Fourier transform can be used to express the signals oscillation amplitude at specific frequency bands regardless of phase. This allows the enhancement of both phase and non phase-locked activities provided they should occur at relatively defined and stable frequencies. However, the Fourier transform contains no temporal information. The Fourier transform can be performed on successive EEG segments defined by a ‘windowing’ function. This procedure allows the extraction of both time and frequency domain information. The windowed Fourier transform and the wavelet transform are such Fourier derived methods. In analogy to Heisenberg’s uncertainty principle, the width of the ‘windowing’ function limits both the time and frequency resolution of the analysis. Windowed Fourier transform uses a fixed and arbitrarily defined window width resulting in a fixed time-frequency resolution ratio. By adapting the window width in function of the frequency, wavelet analysis offers an optimal compromise for timefrequency resolution. Using the wavelet transform of EEG epochs, we explored ERD and ERS induced by noxious and non-noxious CO2 laser stimuli. We hypothesized that these induced rhythms may reflect modulations of cortical networks implicated *
Eventhough ERPs cannot truly be considered as oscillatory, they are often described as
‘phase-locked’ to stimulus onset as opposed to ‘spontaneous’ EEG oscillations which are not.
102
in the central processing of the stimulus. When concomitant activation of A and C-fiber nociceptors occurs, A -fiber mediated transient changes in the state of these networks could modify the electrophysiological correlate of processes induced by the later arriving C-fiber afferent volley. If true, this hypothesis could explain the non-appearance of the C-fiber related ultra-late LEP. 2 Methods 2.1
Subjects
After obtaining informed consent, experiments were performed on 10 healthy volunteers (3 females and 7 males) aged between 22 and 49 years. Subjects were at first familiarized with the experimental surroundings, the noxious stimuli, and the reaction-time acquisition procedure. The rules of the Ethics Committee of the Université Catholique de Louvain Faculty of Medicine were followed. 2.2
Test stimulus and CO2 laser stimulator
Test stimuli were delivered by a CO2 laser designed and built in the Department of Physics at the Université Catholique de Louvain (Plaghki et al. 1994). The CO2 laser system generated a highly collimated infrared beam (wave length: 10.6 µm). The power output was adjustable continuously between 1 and 25 W. Heat pulse duration was 40 ms and beam diameter was 10 mm at the target site. The laser stimulus was highly reproducible (± 1%). The inter-stimulus interval varied randomly between 30 and 40 seconds. To reduce receptor fatigue or sensitization by skin overheating, the target traveled randomly inside a 16 cm2 surface area of skin between the first and second metacarpal bones of the left hand during this interval. To avoid any visual or acoustic clue, all equipment possibly associated with the production of the stimulus was outside the visual field of the subject who also wore headphones as a muffler.
103
2.3
Experimental design
Experimental sessions were divided into four successive blocks. One block consisted of 30 test stimuli of six almost equally spaced intensities ranging between 3.3 and 10.9 mJ/mm2 and repeated five times. Stimulus intensities were delivered in random order. Sessions lasted approximately two hours. 2.4
Data acquisition
2.4.1 Reaction time and intensity of perception A warning buzzer, transmitted through the headphones, signaled the beginning of each trial. The foreperiod between the warning signal and the test stimulus varied at random between 2 and 5 sec (rectangular distribution). Subjects were instructed to push as fast as possible a micro-switch held in the right hand when perceiving the laser stimulus. RT was measured using a computer clock initialized when the laser shutter was opened and halted when the micro-switch was pressed. If the subject did not detect the stimulus within 2.5 seconds, the clock was automatically stopped. After each trial, subjects were asked whether or not the stimulus was perceived as painful using a 101 point visual-analogue scale (VAS). The middle of the scale (VAS=50) marked the borderline between non-painful and painful domains of sensation. Extremities of the scale were annotated ‘no detection’ and ‘maximum pain’. 2.4.2 Electroencephalogram Subjects were instructed to keep eyes open while waiting for the warning signal. EEG was recorded from 19 Ag-AgCl electrodes placed on the scalp according to the International 10-20 system and referenced to linked earlobes. Impedance was kept below 5 kOhm. Ground was placed at the right wrist. Two electrodes, one placed at the upper left and the other at the lower right side of the right eye, monitored ocular movements and eye blinks. EEG was amplified (gain: 1000; filter: 0.06-75 Hz) using a paperless-EEG recorder (PLEEG, Walter Graphtek, Germany). For the purpose of this study, three channels of interest (electrodes CZ, PZ and C4) were sampled at 256 cps via a 12-bit A/D converter CED 1401 (Cambridge Electronics Design, UK). Epochs
104
extended from 500 ms before to 3500 ms after stimulus onset (1024 points). After baseline-correction (reference interval –500 to 0 ms), sweeps contaminated by EOG were rejected by visual inspection. 2.5
Data analysis
EEG epochs were classified according to stimulus intensity resulting in 6 categories. For the reasons exposed in the introduction, RT was used to classify EEG epochs in three groups. The first group (‘Late RT’) consisted of detected trials with RTs ≤ 650 ms. The second group (‘Ultra-Late RT’) consisted of detected trials with RTs > 650 ms. As shown in figure 3-1, this cut-off value optimally separated the bimodal distribution of RTs. The third group (‘No Detection’) consisted of undetected trials. 2.5.1 Time-frequency transformation of the data A time-frequency (TF) representation based on the continuous Morlet wavelet transform (CMT) of EEG epochs was used to identify stimulus-induced amplitude modulations of oscillatory activities*. Explored frequencies ranged from 1 to 45 Hz in steps of 0.22 Hz. ‘TF-single’ transform: To enhance EEG changes time-locked but not necessarily phase-locked to stimulus onset (i.e. ERD and ERS but also ERPs), the CMT was applied to each individual trials. Resulting TF amplitude maps were then averaged across trials for each subject and within each group. These maps (labelled ‘TF-single’) express the average oscillation amplitude as a function of time and frequency. ‘TF-average’ transform: According to the additive noise model (Regan 1989), time-averaging will only enhance stimulus-related EEG changes whose waveform is identical across trials (i.e. ERPs). Stimulus related oscillations satisfy this condition only if phase-locked to the stimulus onset. The TF
*
− t2 The relative spread of the mother Morlet function ψ (t ) = exp exp( jω0t ) was set to 2σ ²
σ = 2.5 πω0
(see Appendix A for methodological details).
105
transform of time-averaged trials (labelled ‘TF-average’) was computed for each subject and group. The ‘TF-single’ transform thus displays time-locked EEG changes whether or not phase-locked to stimulus onset while the ‘TF-average’ transform displays time-locked EEG changes only if phase-locked to stimulus onset. To display results as a relative increase or decrease of oscillation amplitude as compared to a pre-stimulus baseline level (i.e. -350 to -150 ms)*, z-scores were computed for each group on the grand-average across subjects (figure 3-3). This procedure was performed separately for each computed frequency lines. All computations were carried out by software developed in Delphi v6.0 (Borland Software Corporation, USA). 3 Results 3.1
Intensity of perception and reaction time
At the lowest stimulus intensity, 15% of trials were not perceived. At these and intermediate intensities, a sensation of diffuse warmth was most often reported. At higher intensities, subjects often reported an additional sharp pricking sensation. On average, 74 ± 12% of the applied laser stimuli were detected and 39 ± 22% of detected trials were rated as painful (VAS ≥ 50). As expected, frequency distribution of RTs was bimodal. As shown in figure 31, a cut-off value of 650 ms correctly discriminated two groups of RTs, labeled ‘Late RT’ ( 650 ms; n = 456) and ‘Ultra-Late RT’ (> 650ms; n = 389). The proportion of ‘Late RTs’ clearly increased with the intensity of the laser stimulus. The group of undetected stimuli (‘No Detection’) consisted of 355 trials.
*
At lower frequencies, the wavelet function and subsequently its time resolution is very
spread out in the time domain. To reduce the participation of post-stimulus oscillation amplitudes, the time-interval extending from -150 to 0 ms was not included in the reference period. For the same reason, the time-interval extending from -400 to -350 ms was rejected to reduce the participation of artefacts induced by zero-padding.
106
Late RT
650 ms
Ultra-late RT 35% 0%
10.9 9.1 7.3 5.7 4.3
0
500
1000
Reaction times (bins 100 ms)
1500
3.3
Fig. 3-1. Contour map displaying reaction time (RT) distributions (10 subjects, 120 stimuli per subject) as a function of CO2 laser stimulus strength (six intensities ranging from 3.3 to 10.9 mJ/mm2). Distribution of RTs was clearly bimodal. A 650 ms cutoff time value correctly separated two groups of RTs. Ultra-late RTs were mostly found at lower stimulus intensities (at 3.3 mJ/mm2, average of RT was 1128 +/- 352 ms) while Late RTs were mostly found at higher stimulus intensities (at 10.9 mJ/mm2, average of RT was 464 +/- 189 ms).
3.2
Evoked potentials
After artefact rejection, 816 out of 1200 recordings remained. Stimuli of high intensity often provoked an eye blink reflex. For this reason, a greater proportion of EOG artefacted recordings was found in group ‘Late RT’ (46%) and ‘Ultra-late RT’ (33%) than in group ‘No Detection’ (12%). Time averaged epochs (one subject displayed as example in figure 3-2) in the ‘Late RT’ group displayed a negative peak (N2) at 210 ± 27 ms followed by a positive peak (P2) at 380 ± 53 ms. Both peaks were maximal at electrode CZ. In the ‘Ultra-late RT’ group, time averaged epochs showed a small positivity at latencies around 1050 ms. In group ‘No Detection’, averaged epochs did not show stimulus-evoked changes in the latency range of late and ultra-late LEPs.
107
CZ
Late RT group N2
P2
Ultra-late RT group
No Detection group
10 µV +
500 ms
Fig. 3-2. Laser-evoked potentials (LEP) recorded at the vertex (CZ v.s. linked earlobes, low-pass filter : 20 Hz) in subject L.M. Time-average (thick line) of ‘Late RT’ group (n=32) displays a late N2 / P2 complex. Time-average (thick line) of ‘Ultra-late RT’ group (n=25) displays a less ample positivity maximal at 1110 ms which is compatible with the latency of the ultra-late C-fibre evoked LEP (750-1150 ms; Bromm et al., 1983; Bragard et al., 1996). Time-average (thick line) of ‘No Detection’ group does not show evoked changes. Even and odd trials were averaged separately and are displayed as thin grey lines.
3.3
Time-frequency analysis
The average of TF transforms of single trials (‘TF-single’ transform) expresses EEG oscillations whether or not phase-locked to stimulus-onset. Left panels of figure 3-3 display, at electrode CZ, stimulus related increases (yellow) and decreases (red) of oscillation amplitude (z-scores) for the ‘No detection’, ‘Ultra-late RT’ and ‘Late RT’ groups. Right panels of figure 3-3 display the TF transforms of time-averaged trials (‘TF-average’ transform), expressing only activities which are ‘phase-locked’ to the stimulus onset. Activities present in both the ‘TF-single’ and the ‘TF-average’ transforms should be considered as phase-locked to stimulus onset (i.e. ERPs) while activities present only in the
108
TF-SINGLE
TF-AVERAGE
40 30 20 10 1 40 30 5
20 2
10 1 40 30 3
20 10 1
4
1 0
1
2 Time (s)
3
0
1
2
3
Time (s) z
- 25
+ 25
Fig. 3-3. Time-frequency estimation of oscillation amplitude was obtained by taking the norm of the Morlet transform of EEG time-series recorded at CZ following CO2 laser stimulation (see Methods for details). Left panels display the average of TF transforms of single trials (‘TF-single’ transform) enhancing both phase and non phase-locked stimulus related EEG changes (grand averages across subjects). Right panels display the TF transform of time-averaged trials (‘TF-average’ transform) enhancing only activities both time and phase locked (grand averages across subjects). Comparison between left and right panels allows distinguishing between phase and non phase-locked activities related to laser stimulus in groups ‘Late RT’, ‘Ultra-late RT’ and ‘No detection’. Results are displayed as increases and decreases of oscillation amplitude (zscores), relative to a reference period (-350 to -150 ms). Suspected foci of stimulusrelated EEG changes, circumscribed by dashed lines, were then enclosed in arbitrarily defined regions of interest (numbered 1-5, see figure 3-5).
109
‘TF-single’ transform should be considered as non phase-locked to stimulus onset (i.e. ERD and ERS).
CZ
Fig. 3-4. Time-average (thick black line) of all EEG epochs (CZ vs. linked earlobes) of the Late RT group (n = 456) after bandfiltering (FFT filter) in three different frequency bands (0-5, 5-8, 812 Hz). The unfiltered timeaverage is displayed as a grey dashed line for comparison. Only low frequency components (0 - 5 Hz) participate significantly in the genesis of the late LEP complex. Even and odd trials were averaged separately and are displayed as thin grey lines.
0 - 5 Hz
5 - 8 Hz
8 - 12 Hz 10 µV +
500 ms
For reasons exposed thereafter, five TF regions of interest (ROI 1-5) circumscribing suspected foci of stimulus related EEG modulations were defined as shown in figures 3-5. In addition, ROI 0 was defined in the foreperiod to estimate the non stimulus related variance across groups. Amplitudes contained within each ROI were averaged and then compared between detected (‘Late RT’ or ‘Ultra-late RT’) and ‘No detection’ groups (used as ‘control’) using the non-parametric Wilcoxon test for matched pairs*. This test was applied to the ‘TF-single’ transform at electrode CZ except when specified otherwise. Results are summarized in Table 3-1 and presented in the next sections.
*
Several suspected foci of stimulus-related EEG changes were identified in both the ‘TF-
average’ and the ‘TF-single’ transforms (dashed-lines in figure 2-3). Regions of interest (ROI) circumscribing these suspected activities were arbitrarily defined (figure 2-5). To assess their statistical significance, amplitudes contained within each ROI were averaged for each subject. These were then compared between detected (‘Late RT’ or ‘Ultra-late RT’) and ‘No Detection’ groups used as ‘control’. Since averaged amplitudes showed a far from normal distribution, the non-parametric Wilcoxon test for matched pairs (two-tailed) was preferred. Results are summarized in table 2-1.
110
Late RT group
Ultra-late RT group
(Hz) 30 20
0
0
3
10
5
4 2
1
1 0
1
0
1
2 (s)
Fig. 3-5. Average of TF transforms of Late RT and Ultra-late RT groups were used to arbitrarily define regions of interest (ROI numbered 1 - 5) circumscribing suspected foci of stimulus related increases or decreases of oscillation amplitude (grey-filled contours). ROI 0, defined in the foreperiod, was used to estimate the non stimulus related variance across groups.
3.3.1 Phase-locked evoked changes Two ROI displayed EEG activity clearly present in both the ‘TF-single’ and the ‘TF-average’ transforms. 3.3.1.1
ROI 1 or ‘Late LEP’
A first focus of phase-locked amplitude increase could be identified in both types of TF representations of the ‘Late RT’ group (figure 3-3 lower panels). In the ‘TF-average’ transform, the frequency spectrum of this activity ranged between 1 and 5 Hz. The ‘TF-single’ transform showed however a much wider enhancement extending beyond 15 Hz. To determine whether these higher frequency components participated in the genesis of the time-averaged late LEP, trials of the ‘Late RT’ group were filtered (FFT filter) in three different frequency bands (0-5 Hz, 5-8 Hz and 8-12 Hz) and averaged. As shown in figure 3-4, this procedure indicates that only low frequency activities contribute significantly to the averaged late LEP. The amplitude increase seen at frequencies above 5 Hz thus mainly reflects a distinct, mostly non phaselocked, activity which was studied separately (ROI 3). To study this first focus of phase-locked activity, a window extending from 130 to 500 ms and 1 to 5 Hz was defined (figure 3-5: ROI 1). Within this region, amplitudes were on average maximal at 296 ms and 3.8 Hz. This activity was 111
present (p < 0.005) in both TF transforms of all three recording electrodes and increased with stimulus strength (figure 3-7). A significant increase was however also found in both the ‘TF-single’ (p
