Nociceptive processing in the human central nervous system can be recorded in different ways, including through the use of hemodynamic and neuroelectric methods.Reference Apkarian, Bushnell, Treede and Zubieta 1 - Reference Peyron, Laurent and Garcia-Larrea 4 Recently, high temporal resolution electrophysiological recordings have become popular as a way of better understanding when painful afferent volleys to the cortex are processed and how these relate to the psychophysical properties of experienced pain.Reference Shao, Shen, Yu, Wilder-Smith and Li 5 Human electroencephalographic (EEG) recording studies together with source localization analysis of generators have: i) confirmed the presence and sub-second activation pattern of brain structures tied to pain intensity processing,Reference Perchet, Godinho and Mazza 6 ii) shown that individual differences in pain sensitivity partly depend on individual differences in the early activation response of the precuneus,Reference Goffaux, Girard-Tremblay, Marchand, Daigle and Whittingstall 7 iii) demonstrated that the primary somatosensory cortex, when activated soon after a painful stimulus, does indeed participate in the processing of painful experiences,Reference Nir, Lev, Moont, Granovsky, Sprecher and Yarnitsky 8 and, iv) shown that regulatory areas of the orbitofrontal cortex and amygdala participate in the initial activation and control of descending bulbo-spinal pain inhibition circuits.Reference Moont, Crispel, Lev, Pud and Yarnitsky 9
Although these studies are useful in their ability to highlight how the human brain responds to painful experiences, they have only considered neuronal networks which oscillate in phase to painful afferent volleys. This type of activity known as evoked - or phase locked - can be instructive regarding the role of synchronization in post-stimulus EEG amplitude changes (i.e., post-stimulus reorganization of oscillatory neuronal networks). However part of the brain's response to an afferent sensory stimulus is not rigidly phase locked to stimulus onset but, instead, will vary slightly from trial-to-trial. Bursts of such asynchronous oscillatory activity are called induced – or non-phase locked – responses. These have very rarely been explored in pain research. Evoked and induced brain responses are thought to represent different neurophysiological brain processes. Evoked responses, for example, are thought to reflect the resonant activity of cortical neurons in response to incoming sensory information.Reference Pfurtscheller and Lopes da Silva 10 Induced responses, on the other hand, are thought to reflect the indirect synchronous firing of neuronal assemblies produced by large fluctuations in the membrane potential of neurons.Reference David, Kilner and Friston 11 These fluctuations occur close to, and depend on, the coherent activation patterns normally triggered by incoming sensory information.
Few studies have investigated induced EEG response in pain. The first is a study by Mouraux et al.,Reference Mouraux, Guerit and Plaghki 12 which tested for the possible interaction between Aδ and C fiber-evoked EEG signatures to pain. Using both phase and non-phase locked measurements of Aδ and C fiber-mediated responses to pain (uncoupled using laser stimuli and reaction time tasks), the authors showed that Aδ and C fibers produce the same brain response, which, when already engaged by rapidly-travelling Aδ fibers, do not need to be reactivated by slowly-travelling C fibers. The second key paper was published by Babiloni et al.Reference Babiloni, Babiloni and Carducci 13 and showed that EEG oscillatory responses (both phase and non-phase locked) to a painful stimulation occur in both hemispheres. Although the EEG response recorded by Babiloni et al.Reference Babiloni, Babiloni and Carducci 13 was dominated by phase locked activity, a significant non-phase locked burst of alpha activity was observed. More importantly, it was shown that, depending on the frequency of interest, phase and non-phase locked EEG oscillations have different scalp topographies and, thus, could be related to different underlying sources and involve different pain processing functions. Despite being important to the field of pain neuroimaging, these two papers did not seek to address the functional significance of non-phase locked EEG responses to pain. To address this issue, we analysed EEG responses to pain in the time-frequency domain, source-localized underlying brain generators and explored how these changes relate to the psychophysical properties of pain.
Methods
Subjects
Thirteen women (mean age±standard error, 23.8±0.65 years) and ten men (mean age±standard error, 26.3±2.03 years) participated in the experiment. The study was carefully explained to all participants and all provided written, informed consent. The study was approved by the research ethics committee of the Centre Hospitalier Universitaire de Sherbrooke (CHUS).
Subjective ratings
Verbal numerical rating scales (NRS) were used to evaluate the intensity of all somesthetic sensations. An innocuous NRS was used to evaluate non-painful sensations and ranged from 0-100, where 0 was defined as “no sensation” and 100 was defined as “extremely intense, but not painful”. A noxious NRS was used to evaluate painful sensations and also ranged from 0-100, but this time, 0 was defined as “no pain” and 100 was defined as “intolerable pain intensity”. In order to distinguish between the scales, participants had to precede all innocuous evaluations by the word “non-painful”. Using separate 0-100 scales to assess non-painful and painful sensations respectively was preferred to the use of a single 0-200 scale (where 100 would have represented the pain perception threshold) because it was feared that a single 0-200 scale might have been less intuitively used by participants, and thus, might have led to important evaluation errors.
Sural nerve stimulations
Both non-painful and painful sensations were provoked using transcutaneous electrical stimulations of the right sural nerve. The sural nerve was stimulated over its retromalleolar path. Stimulations consisted of a volley of 10 electrical pulses (square waves - 1 ms pulse duration) administered at a rate of 320 Hz using a constant current stimulator. A stimulation volley lasted 31 ms. Stimulations were provided using a pseudorandom inter stimulus interval of 6-12 s (geometrical distribution with a mean of 7.4 stimuli per minute). The use of a jittered design allowed us to control for pain-expectation effects, which are known to influence subjective pain reportsReference Carlsson, Andersson, Petrovic, Petersson, Ohman and Ingvar 14 . Stimulations were provided in two separate testing blocks. Each block lasted 4.5 minutes with an inter-block interval of 5 minutes. The first block contained 33 stimuli provided at a stimulation intensity level necessary to provoke strong tactile (but non-painful) sensations and corresponding to a score of 50 on the innocuous NRS (individually tailored). The second testing block also contained 33 stimuli but this time the intensity level was adjusted to ensure the subjective experience of mild pain - corresponding to a score of 15 on the noxious NRS. Mild, as opposed to moderate or elevated pain was targeted in our study because sural nerve sensations are typically experienced as more unpleasant than intense by participants. As a result, high levels of targeted pain intensity run the risk of causing extremely high levels of pain unpleasantness and thus elevated drop-out rates. Stimulation intensity always remained constant within blocks. The non-painful testing block was always presented before the painful testing block. This is important because painful testing blocks can potentially produce spinal sensitizing effects that can carry-over and affect the evaluation of all subsequent blocks. The use of an incremental testing design, therefore, prevented potential carry-over effects from the painful to the non-painful block.
Somatosensory-evoked brain potentials
Shock-evoked somatosensory-evoked potentials (SEPs) were recorded from 32 electroencephalographic electrodes arranged on the scalp in a grid corresponding to the international 10–20 system (32-electrode cap). Using a Brain Vision recording system (Brain Products GmBh, Munich, Germany), our signal was sampled at 250 Hz and bandpass filtered between 0.5 to 30 Hz. A common average reference montage was used. Impedance at each electrode position was kept below 5 kΩ. Trials contaminated by eye-blinks and movements were corrected using an independent component analysis (ICA) algorithm (Brain Vision Analyzer software, Brain Products GmbH, Munich, Germany). After ICA, an additional baseline correction (−100 to 0 ms, where 0ms represents shock onset) was applied. Each trial spanned a −100 ms to 600 ms time period.
Time-frequency analysis
Time-frequency (TF) analysis was run using a continuous Morlet wavelet transform (CMT). The CMT was computed using absolute values of the raw EEG data and was baseline corrected. Our CMT baseline correction method was performed by calculating the average value of the wavelet coefficients between −100 ms to 0 ms and subtracting this value from all wavelet coefficients within the entire epoch (i.e., −100 ms to 600 ms) and for all corresponding frequency levels (1 Hz resolution). A subtraction approach to baseline correction was preferred to a percentage approach (which calculates a percentage change value from baseline) because the subtraction approach minimizes overestimations in synchronized brain responses.Reference Hu, Xiao, Zhang, Mouraux and Iannetti 15 Following baseline correction, we normalized our wavelet functions using a relative wavelet power method, which ensured that our wavelet function would have the same energy across all scales. This approach produces wavelet power spectra that are directly comparable to each other across all scales. Time-frequency analyses were applied only to average-removed single trials, which isolated the induced (i.e., non-phase locked) component of the SEP response.Reference Kalcher and Pfurtscheller 16 Time-frequency analyses were conducted separately on all trials within painful and non-painful blocks.
Cortical mapping
Visual inspection of the time frequency oscillation amplitude maps depicted in Figure 1 clearly show a pain minus non-pain burst of delta/theta activity spanning a 180 to 500 ms time period and located across frontal and posterior leads (Note: the difference in delta/theta scalp activity between pain and non-pain trials was statistically significant; see Results section below). To properly investigate the underlying source of this non-phase locked activity, we first segmented the pain and non-pain raw EEG data between 180 and 500 ms. The segmented data were individually transformed into a cross-spectrum file with a discrete frequency range extending between 1 and 30 Hz and a frequency resolution of 1.95 Hz (this was the upper limit of our resolution given our segment length of 320 ms [i.e., 180 ms to 500 ms]). Following the computation of both pain and non-pain cross spectrum files (completed separately for every participant), a source file was computed to estimate the source generators at each discrete frequency (15 discrete frequencies in total). Non-phase locked data were transformed from scalp space to source space using standardized low-resolution electromagnetic tomography (sLORETA). Standardized low-resolution electromagnetic tomography is a method to estimate the localization of brain function by providing a solution to the inverse problem.Reference Pascual-Marqui, Esslen, Kochi and Lehmann 17 , Reference Pascual-Marqui, Michel and Lehmann 18 It falls into the class of distributed source models that do not require a priori knowledge regarding the number of active sources and thus is advantageous for the localization of processes that depend on multiple activity distributions, as is the case for pain.Reference Apkarian, Bushnell, Treede and Zubieta 1 , Reference Schweinhardt and Bushnell 19 , Reference Tracey and Mantyh 20 The source space in sLORETA is calculated with the boundary element method applied to the standardized Montreal Neurological Institute (MNI) brain volume consisting of 6239 cortical gray matter voxels at five-mm resolution. In the present study, the computation of sLORETA (which provides electric current density values) was applied, for each participant, to the non-phase locked data obtained at each of our 15 discrete frequencies on both non-pain and pain trials.
Figure 1 Time-frequency decomposition of non-phase locked oscillatory activity recorded at electrode Oz (distribution across the entire scalp is shown in upper right corner). Time frequency data shown here represents the activity difference between pain and non-pain trials. Estimations were obtained using a continuous Morlet transform applied to average-removed single trials. An event-related synchronization response is clearly visible between 180 and 500 ms and ranging between 1.5 and 7 Hz.
Experimental protocol
Prior to testing, a pre-experimental session was carried out to individually determine the current intensities to be used during the testing session. Participants were asked to rate the strength of the sensation they felt following each practice stimulation during the pre-experimental session. This allowed us to determine the current intensity in milliamperes (mA) required to elicit both non-painful (i.e., 50 on the innocuous NRS) and mildly painful (i.e., 15 on the noxious NRS) sensations for each participant. It also allowed us to determine the pain perception threshold (in mA). During testing, intensity ratings were recorded three times during the non-painful block and three times during the painful block. For each block, intensity ratings were obtained once every 1.5 minutes and reflected the average pain felt since the beginning of a block (for the first assessment) or since the preceding assessment (for the last two assessments). Having three scores within a given block allowed us to test for the stability of ratings within blocks. It also allowed us to calculate the difference between the last and first score within a block, giving us an opportunity to explore inter-individual variations in subjective rating as testing time progresses.
Statistical analyses
Subjective intensity ratings and sural nerve stimulation strength
To ensure that the average intensity rating within a given block corresponded to the intensity value that was targeted for that block, we conducted two separate one-sample t-tests (one each for the non-painful and painful testing block). A two-way repeated measures analysis of variance (ANOVA) (Block X Rating Occasion) was also conducted to explore possible changes in subjective rating within testing blocks (i.e., rating changes through time). Bonferroni corrections were applied where appropriate.
Pain-related cortical activation maps associated with non-phase locked SEP responses
Cortical activity tied specifically to pain induced responses were estimated by comparing the electric current density maps obtained from the non-phase locked data in the non-pain condition from the current density maps obtained from the non-phase locked data in the pain condition. This comparison was made at each of our 15 discrete frequencies. Statistical significance was assessed through nonparametric permutation tests using 5000 randomizations accounting for multiple comparisons (i.e., multiple anatomical sites and multiple frequency intervals). Significance was set to p<0.05 (two-tailed tests were always used). Statistical non-parametric mapping provided 3D t-maps for all consecutive frequencies in the 180 to 500 ms post-shock epoch.
Correlation analysis
To explore the functional significance of pain induced responses, we correlated various psychophysical pain indices (i.e., pain threshold, sub- and supra- pain-threshold scores, pain sensitivity scores and change in subjective experience as stimulations repeat through time) with the source activity (in amperes per square meter, (A/m2)) of pain-induced brain responses. The pain-related brain regions chosen for these correlations were limited to those regions (spatial local maxima) which were significantly different when pain and non-pain induced responses were compared (see description of this analysis in the preceding section). The psychophysical pain indices chosen in the current study were the pain threshold (in mA), the sub- and supra- pain-threshold (both in mAs), the pain sensitivity index (change in sural nerve stimulation current between the non-painful and painful blocks), and the change in the subjective rating between the last and first score within a block (both non-painful and painful, each calculated separately). Bonferroni corrections were applied to limit the risk of type I error.
Results
Subjective ratings
Average NRS ratings for the non-painful (mean=53.40±standard error (SE)=3.70) and painful testing blocks (mean=15.10±SE=1.75) were compared to their corresponding target NRS values (50 and 15, respectively). Numerical rating scale ratings for the non-painful testing block (t22=0.92, p=0.37) and for the painful testing block (t22=0.058, p=0.95) were not significantly different from their targeted values. For both testing blocks, intra-block subjective ratings (i.e., first, second, and third ratings) did not vary significantly from one another (F2, 44=2.81, p=0.07; see Table 1).
Table 1 Subjective intensity ratings and sural nerve stimulation strength (N=23).
N=number; mA=milliampere
Source localization of non-phase locked SEP responses
Visual inspection of the time frequency oscillation amplitude map depicting the difference between painful and non-painful stimulations clearly shows a burst of delta/theta activity spanning a 180 to 500 ms time period (See Figure 1) and localized across both frontal and posterior leads. This burst of localized activity resulted from significantly greater delta and theta power on pain trials than on non-pain trials at frontal leads (peak activity at electrode Fp1, t22=2.55, p=0.02) and a marginal difference at posterior leads (peak activity at electrode Oz t22=1.74, p=0.09). Within the 2.9 to 4.9 Hz range, generators for this burst of induced, pain-related activity were localised to the left superior frontal gyrus (SFG; spatial local maximum at x=−20, y=30, z=50; t22=3.75, p<0.05) and right middle frontal gyrus (MFG; spatial local maximum at x=50, y=15, z=45; t22=3.83, p<0.05; see Figure 2a). Within the 4.9 to 6.9 Hz range, generators were localized to the left (spatial local maximum at x=−25, y=30, z=55; t22=3.75, p<0.05) and right SFG (spatial local maximum at x=25, y=25, z=55; t22=3.61, p<0.05), and, to the right MFG (spatial local maximum at x=45, y=20, z=40; t22=3.68, p<0.05; see Figure 2b).
Figure 2 Coronal views (y=30) showing the current density differences between painful and non-painful conditions. Significant differences were located within the left superior (1) and right middle (2) frontal gyrus in the 2.9 to 4.9 Hz frequency range (see a) and within the left (3) and right (4) superior frontal gyrus and in the right middle (5) frontal gyrus in the 4.9 to 6.9 Hz frequency range (see b). All local maxima differences (white dots) were significant. Color bar at bottom represents positive (red) and negative (blue) t-statistic values.
Correlation analyses
The only significant association between induced brain activity and psychophysical pain measures implicated the right MFG within the 4.9 to 6.9 Hz range (x=45, y=20, z=40) and involved only the change in subjective pain experienced as testing progressed through time (r=−0.56, p=0.006; see Figure 3). This means that pain-related increases in induced theta activity within the right MFG were associated with a decrease in experienced pain. No other correlation coefficient was significant (all r values (rs) less than |0.44|, all p values (ps) less than 0.05; p values here always include the application of Bonferroni corrections).
Figure 3 Induced theta brain activity obtained during the pain block and located in the right middle frontal gyrus correlated negatively with the change in subjective pain intensity (i.e., the difference between the last and first pain perception score within the pain block).
It is important to point out that Hu et al.Reference Hu, Xiao, Zhang, Mouraux and Iannetti 15 recently conducted a study where they showed that post-stimulus EEG power is partly affected by the trial-to-trial variability in pre-stimulus EEG power. They also found that pre-stimulus EEG power varies following a hyperbolic function of the trial order (i.e., smaller power at the earliest trials followed by larger, relatively stable power responses thereafter). To test for the presence of trial-to-trial variability in pre-stimulus power and thus for the possibility of having to correct for pre-stimulus effects, we calculated theta power (via a fast-Fourier transform) within a −100 to 0 ms pre-stimulus time window for all pain trials and compared the average power obtained within the first three trials to the average power obtained in the middle three trials (trials 16, 17 and 18) and last three trials. This was done for power values extracted from electrode Oz, where theta power in the pain block was strongest. Results from the one-way repeated measures ANOVA conducted to test for this effect showed that trial order had no influence on pre-stimulus power (F2, 44=0.08, p=0.93; see Figure 4). As a result, no correction for pre-stimulus EEG power was required or carried out.
Figure 4 Average pre-stimulus theta power (between −100 to 0 ms) calculated for each trial within the pain block. A significant trial effect on the modulation of pre-stimulus power values is not observed.
Specificity of non-phase locked SEP responses
Although our TF analyses were applied only to average-removed single trials, this approach does not guarantee that induced (i.e., non-phase locked) components have been conclusively isolated. This is because subtracting an average response cannot completely eliminate an effect which varies from trial-to-trial. Thus, to ensure that our induced effects are indeed specific (or at least not dominated by evoked effects), we re-ran our analyses using only evoked data (i.e., data obtained from single subject average responses). Since our objective here was to confirm the specificity of our induced results and not to chart the full spectrum of synchronous and asynchronous pain-evoked responses, we limited our analyses to the post-shock epoch when our induced effects were strongest (i.e., between 180 and 500 ms). Visual inspection of the evoked time frequency oscillation amplitude map (Figure 5) shows greater synchronous activity for pain trials than for non-pain across multiple scalp regions and frequencies. Pain versus non-pain differences calculated specifically across frontal and posterior leads and specifically for delta/theta frequencies were significant at frontal leads (peak activity at electrode Fp1, t22=2.63, p=0.01) and marginal at posterior leads (peak activity at electrode Oz t22=1.76, p=0.09). This is pattern of activity is comparable to the pattern obtained when analyzing induced responses (see description of induced effects above). Source generators, however, were not the same as those obtained for induced responses. Within the 2.9 to 4.9 Hz range, generators for evoked responses were localised to the right rectal gyrus (spatial local maximum at x=10, y=35, z=−25; t22=5.30, p<0.001; see Figure 6a). Within the 4.9 to 6.9 Hz range, generators were localized to the right inferior frontal gyrus (spatial local maximum at x=50, y=40, z=15; t22=4.28, p<0.05; see Figure 6b). Neither one of these generators were associated with our psychophysical pain measures (all rs<|0.42|, all ps>0.05; p values here include the application of Bonferroni corrections). This suggests that the induced effects reported earlier are not dominated (or driven) by evoked effects.
Figure 5 Time-frequency decomposition of phase locked (i.e., evoked) oscillatory activity recorded at electrode Oz (distribution across the entire scalp is shown in upper right corner). Time frequency data shown here represents the activity difference between pain and non-pain averages (across participants). Estimations were obtained using a continuous Morlet transform applied to both pain and non-pain averages.
Figure 6 Current density differences between painful and non-painful conditions were significant within the right rectal gyrus in the 2.9 to 4.9 Hz frequency range (see a for an axial view at z=−25) and within the right inferior frontal gyrus in the 4.9 to 6.9 Hz frequency range (see b for a coronal view at y=40). All local maxima differences (white dots) were significant. Color bar at bottom represents positive (red) and negative (blue) t-statistic values.
Discussion
The purpose of this study was to investigate the functional significance of non-phase locked (induced) brain responses following painful events. Results showed that following a painful electrical stimulation of the right sural nerve, a burst of delta/theta activity occurred soon after shock onset (between 180 and 500 ms). This activity was source-localized to the middle and superior frontal gyri (bilaterally). Importantly, increased theta activity within the right MFG predicted decreased sensations of pain as stimulations repeated within the pain block. The MFG is an anatomical region which makes up part of the functional structure known as the dorsolateral prefrontal cortex (DLPFC). Given the association found here between DLPFC activity and subjective pain, our results suggest that the DLPFC may be associated with processes related to the attenuation of repeating painful experiences.
Results reported here agree with those of a previous positron emission tomography study conducted by Lorenz et al.Reference Lorenz, Minoshima and Casey 21 which showed that the association between the anterior insula (a well-known pain processing node in the human brain) and subjective pain intensity scores decrease when right DLPFC activity is high. This suggests that the DLPFC exerts active control of pain perception by modulating cortico-cortical pathways involved in the processing of nociceptive afferents. Although the Lorenz et al. paper constitutes a seminal study on the role of the prefrontal cortex in pain, the neuroimaging technique used depends on haemodynamic brain responses – which are notoriously poor at capturing rapid fluctuations in post-shock brain activity (but well suited to capture three dimensional sources). Haemodynamic-based neuroimaging data also fail to provide a direct measure of neuronal activity, let alone offer results which can distinguish between evoked and induced neuronal brain responses. In the current study, the DLPFC activity observed was induced (i.e., devoid of any contribution from the phase-locked average response) and occurred soon after shock onset, which can be instructive regarding the functional significance of this response. Variations in induced brain activity have been proposed to reflect functional changes in the parameters controlling dynamic interactions within and between brain structures (i.e. a changes in the synchrony of underlying neuronal populationsReference Alba, Marroquin, Pena, Harmony and Gonzalez-Frankenberger 22 ). This means that our rapid post-stimulus DLPFC response may represent a control process which relates to physiological connectivity in the cortex and may thus represent a measure of long-range interaction between brain structures (see Alba et al.Reference Alba, Marroquin, Pena, Harmony and Gonzalez-Frankenberger 22 for an original formulation of this idea). This likely explains why, in our study, the DLPFC was associated with a progressive decrease in pain intensity (reflecting the exercise of control) rather than with baseline pain sensitivity (reflecting an inceptive response to pain).
In the current study, some limitations are worth mentioning. First, the association we found between pain-induced brain activity and the change in pain observed as testing advanced was based on the brain activity recorded and averaged across the entire pain block. As a result, we do not actually measure change in brain activity between the earliest and latest fractions of the pain block. The current testing design was simply not intended to partial out early and late block effects in the EEG response (i.e., too few trials within a block to compare brain responses to the first and last trials). Another limitation of the current study is that we do not know if the activation of the right DLPFC represents a lateralized or a right brain effect. To address this question, a testing design where both right and left sural nerve shocks are provided would be necessary. Moreover, our study cannot characterise the causal effects tied to pain-induced DLPFC responses. That is, we cannot be sure if increased DLPFC activity leads to a change in perceived pain or if a change in perceived pain leads to increased DLPFC activity. To identify causal effects related to DLPFC activity, future studies should use repetitive transcranial magnetic stimulation (rTMS), since this type of stimulation produces direct changes in underlying brain activity. Depending on rTMS frequency, it is possible to either facilitate (at high frequencies) or impede (at low frequencies) synaptic efficiency, which helps pinpoint how a given brain region relates to observed behaviour. Interestingly, a few recent pain studies have used rTMS technology to identify the role played by the DLPFC.Reference Borckardt, Smith and Reeves 23 - Reference Nahmias, Debes, de Andrade, Mhalla and Bouhassira 27 For example, Nahmias et al.Reference Nahmias, Debes, de Andrade, Mhalla and Bouhassira 27 showed that unilateral, high frequency rTMS pulses applied to the right DLPFC produced diffuse analgesic effects in healthy human participants exposed to experimental heat and cold pain. This initial result suggests that the DLPFC may, indeed, play a causal role in reducing painful sensations.
Our findings may also have important implications regarding the comprehension of pain control processes in chronic pain patients. This is because (i) the DLPFC is considered to be part of an executive control network normally associated with pain habituation, pain coping, and pain inhibitionReference Buckner, Andrews-Hanna and Schacter 28 , Reference Fox, Snyder, Vincent, Corbetta, Van and Raichle 29 , (ii) chronic pain patients show structural, neurochemical and functional changes in the DLPFCReference Schweinhardt and Bushnell 19 and, finally, (iii) chronic pain patients fail to show adaptive coping responses (including habituation to prolonged pain) when exposed to prolonged experimental stimuli. Reference Smith, Tooley, Montague, Robinson, Cosper and Mullins 30 Interestingly, a relatively recent study also showed that cerebral potentials recorded in response to repetitive noxious stimuli do not habituate (i.e., do not decrease in amplitude) in patients who suffer from fibromyalgia but do in healthy controls.Reference De Tommaso, Federici and Santostasi 31 This neuronal effect was further tied to reduced habituation of experienced pain. The available scientific literature, together with our results suggest that the DLPFC may serve as a promising target area in the development of neuroimaging-based risk models and/or targeted neurostimulation approaches to treatment.
In conclusion, the current study demonstrated that a progressive decrease in perceived pain intensity was associated with a rapidly occurring burst of theta activity in the right DLPFC. Since this activity was out of phase with respect to stimulus onset, it is unlikely to be directly associated with the processing of noxious afferents but, instead, reflect the response of neuronal networks involved in the control of pain. Future studies, therefore, should investigate the connectivity between the DLPFC and well-known pain-related brain nodes, as well as investigate the role played by this structure in the development and/or maintenance of chronic pain.
Disclosures
Raphaël Dufort Rouleau has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant. Lydia Lagrandeur has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant. Kathya Daigle has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant. Dominique Lorrain has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant. Guillaume Léonard has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant. Kevin Whittingstall has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant.Philippe Goffaux has the following disclosures: Fonds de Recherche du Québec – Santé, Researcher, Research grant; Natural Sciences and Engineering Research Council of Canada, Researcher, Research grant.
Acknowledgements
This work was supported by grants from the Fonds de Recherche du Québec – Santé and the Natural Sciences and Engineering Research Council of Canada. The authors report no conflict of interest concerning the materials or methods used in this study or the findings in this paper.
