abnormal thalamocortical activity in patients with complex...

ARTICLE IN PRESS

PAIN�

xxx (2010) xxx–xxx

w w w . e l se v i e r . c o m / l o c a t e / p a i n

Abnormal thalamocortical activity in patients with Complex Regional PainSyndrome (CRPS) Type I

K.D. Walton a, M. Dubois b, R.R. Llinás a,*

a Dept. of Physiology & Neuroscience, New York University School of Medicine, 550 First Ave., New York, NY 10016, USAb Pain Medicine, Dept. of Anesthesiology, New York University School of Medicine, 550 First Ave., New York, NY 10016, USA

a r t i c l e i n f o

Article history:Received 1 September 2009Received in revised form 19 January 2010Accepted 12 February 2010Available online xxxx

Keywords:Complex Regional Pain Syndrome IIBrain mappingThalamo-cortical DysrhythmiaMEGCentral pain

0304-3959/$36.00 � 2010 International Associationdoi:10.1016/j.pain.2010.02.023

* Corresponding author. Tel.: +1 212 263 5415.E-mail address: [email protected] (R.R. L

Please cite this article in press as: Walton KD ePAIN

�(2010), doi:10.1016/j.pain.2010.02.023

a b s t r a c t

Complex Regional Pain Syndrome (CRPS) is a neuropathic disease that presents a continuing challenge interms of pathophysiology, diagnosis, and treatment. Recent studies of neuropathic pain, in both animalsand patients, have established a direct relationship between abnormal thalamic rhythmicity related toThalamo-cortical Dysrhythmia (TCD) and the occurrence of central pain. Here, this relationship has beenexamined using magneto-encephalographic (MEG) imaging in CRPS Type I, characterized by the absenceof nerve lesions. The study addresses spontaneous MEG activity from 13 awake, adult patients (2 men, 11women; age 15–62), with CRPS Type I of one extremity (duration range: 3 months to 10 years) and from13 control subjects. All CRPS I patients demonstrated peaks in power spectrum in the delta (<4 Hz) and/ortheta (4–9 Hz) frequency ranges resulting in a characteristically increased spectral power in those rangeswhen compared to control subjects. The localization of such abnormal activity, implemented using inde-pendent component analysis (ICA) of the sensor data, showed delta and/or theta range activity localizedto the somatosensory cortex corresponding to the pain localization, and to orbitofrontal–temporal corti-ces related to the affective pain perception. Indeed, CRPS Type I patients presented abnormal brain activ-ity typical of TCD, which has both diagnostic value indicating a central origin for this ailment and apotential treatment interest involving pharmacological and electrical stimulation therapies.

� 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

Chronic pain may be due to an ongoing pathological process,such as cancer, or ‘may be initiated or caused by a primary lesionor dysfunction in the nervous system’ [51]. The latter, known as‘‘neuropathic pain”, is particularly difficult to treat. Its pathophys-iology, as that of pain processes in general, remains a matter of ac-tive study with several theoretical frameworks being considered[1,29,57,75].

Our approach is electrophysiological and focuses on the pres-ence of abnormal, low frequency CNS oscillatory activity [35,39].Aberrant, intrinsic electrical activity in the thalamocortical loopsystem has been shown to generate these abnormal rhythms[26,36]. Such conditions, addressed as Thalamo-cortical Dysrhyth-mia (TCD) [39], have been reported in neuropathic pain[26–28,69,70] and other disorders [9,26,36,67,77]. Here, we pres-ent the results indicating the likelihood that Complex RegionalPain Syndrome Type I (CRPS I), a chronic, neurogenic disorder ofCNS origin [23,24], may share some characteristics with TCD.

for the Study of Pain. Published by

linás).

t al. Abnormal thalamocortical

The diagnostic criteria for CRPS have evolved recently from anon-standardized and idiosyncratically diagnosed set of unrelatedmaladies to a well-defined clinical syndrome [21]. The latest diag-nostic criteria require pain that is disproportionate to the precipi-tating event and should coexist with a set of related sensoryabnormalities (allodynia or hyperalgesia), edema, autonomic dys-function, motor symptoms and/or trophic changes [21]. Based onthe injury type, two classes of CRPS are recognized by the Interna-tional Association for the Study of Pain (IASP): CRPS I where painoccurs in the absence of nerve lesion, and CRPS II where obviousnerve damage is recognized.

Partly due to diagnostic variations, the CNS pathophysiologyunderlying this set of disorders has been investigated only recently[5]. Functional imaging techniques provide clear data concerningCNS alterations, particularly in CRPS I, as elegantly reviewed re-cently [71]. These studies have shown asymmetries in evokedCNS electrical activity and a functional reorganization of somato-sensory and motor brain regions. An important element in thesestudies concerns the possibility of establishing an objective causalrelation of such changes to the subjective clinical findings. Suchdifferences have been directly related to hyperalgesia surface area[43,59] and to the degree of sensory impairment [58] but not tomotor symptoms [41], or to use reduction [42]. The

Elsevier B.V. All rights reserved.

activity in patients with Complex Regional Pain Syndrome (CRPS) Type I.

http://dx.doi.org/10.1016/j.pain.2010.02.023

mailto:[email protected]

http://www.elsevier.com/locate/pain


Table 1Patient demographics and clinical characteristics.

Session Age Sex Medications 1� Pain localization(original injury site)

Duration(months)

P15 19 F A, AC, M, O Lf leg 7P16 15 F A, AC, N, O Lf foot 4P24 23 F A, AC, M, N Lf leg (Lf leg) 11P25 50 M AC Lf hand 120P29 48 M A, AC, O Rt leg (Rt knee) 14P30 25 F A, AC, O Rt leg (Rt ankle) 46P34 62 F N, O Lf foot and ankle (Lf foot) 30P35 17 F A, AC, N Bi lower back to toes

(Lf knee)48

P37 38 F A, AC, N Bi ankle and foot(Bi ankles)

7

P38 17 F AC, M, N Lf arm, chest, neck andupper back (Lf arm)

10

P44 61 F AC, N, O Lf foot (Lf foot) 3P45 54 F A, AC, O Rt forearm and wrist

(Rt shoulder)8

P47 44 F A, AC, N, O Rt lower leg (Rt knee) 27

A, anti-depressants; AC, anti-convulsant; M, muscle relaxant; N, NSAIDS; O, opiates;Bi, bilateral; Lf, left; Rt, right.

2 K.D. Walton et al. / PAIN�


ARTICLE IN PRESS

electrophysiological findings have been correlated with the pres-ence and overall intensity of chronic pain but not with pain inten-sity at the time of the recording [30,43,58,60]. Further, both MEGand fMRI response abnormalities seen in CRPS I correlate well withthe degree of therapeutic effectiveness [44,49,55,59].

Diagnosis requires that the observed CNS abnormalities be re-lated to the chronic state and that the underlying neuronal mech-anisms be present and active during the resting state as well asunder conditions of sensory stimulation or movement. Here, weexamine the possibility that abnormal spontaneous brain activity,in the resting state, may be a common denominator in these pa-tients and thus provide direct insight into the underlying CNSabnormalities. Some results have appeared in the abstract form[11,18].

2. Methods

2.1. Participants

The database for this study comprised 11 women and 2 men(36.4 ± 4.9 years) diagnosed with CRPS I and an equal number ofhealthy women and men (35.4 ± 3.5 years). All patients referredto us had upper or lower extremity disease and met the revisedIASP diagnostic criteria for CRPS I [21]. They all are presented with(1) A precipitating event that was disproportionate according tothe present clinical condition. (2) The presence of spontaneouspain and mechanical allodynia. (3) The presence of trophic changessuch as edema, change in skin temperature, and sudomotor abnor-mality in the distal part of the affected limb. (4) Exclusion of otherdiagnoses.

The NYU Institutional Review Board approved the study and aninformed written consent was obtained from all subjects beforethe MEG recordings. During the MEG recordings, all patients re-ported spontaneous pain and there was no elicited allodynia.

2.2. MEG data acquisition

All MEG recordings were carried out in a mu-metal magneti-cally shielded room. A set of three MEG recordings, each lasting7 min, was made from each subject. Two sets of recordings weremade with the eyes closed (EC) and one with the eyes open (EO).The subjects were asked to relax, but to stay awake. The locationof the head was monitored during the recordings using electrodesplaced at three standard fiducial marker points (left and right pre-auricular points, and the nasion). The head shape, including thelocation of the three fiducial markers, was obtained for each sub-ject using a 3D tracking system by moving a stylus to each fiducialpoint and over the surface of the head (Fastrak, Polhemus, Colches-ter, VT).

MEG recordings were carried out using one of the two wholehead magnetometer systems. One instrument had 148 channels(4D Neuroimaging) and recordings were made while the subjectwas lying down (6 patients P15–P30 in Table 1). The other MEGhad 275 channels (CTF Systems) and the subject sat in an uprightposition (7 patients P34–P47 in Table 1 and all 13 controls).

2.3. MEG data analysis of power spectra

Recordings made with the 4D instrument were corrected forECG and other artifacts following our published protocols [70].Such artifacts and distant noise for recordings made with the CTFinstrument were reduced using a 3rd order gradient [50]. The noiseof the CTF instrument was recorded before each session. Only sen-sors with noise below 10 fT at P2.5 Hz were accepted and includedin the analysis. All raw data were inspected visually and movement

Please cite this article in press as: Walton KD et al. Abnormal thalamocorticalPAIN

�(2010), doi:10.1016/j.pain.2010.02.023

artifacts or periods of large amplitude slow waves were eliminatedfrom further analysis.

Since we were interested in finding if abnormal brain rhythmswere present in patients with CRPS I, the energy of the MEG signalas a function of frequency, the power spectral density (includingdata from all the accepted sensors) was calculated for each record-ing session. We used a multi-taper approach, which provides re-duced-variance calculations of frequency [39,52]. The powerspectrum was normalized at 1 Hz to eliminate the differences inbaseline power that can change according to how close the headis to the imaging sensors.

To quantifying the relative spectral power distribution betweengroups, the mean spectral energy (MSE) [47] was calculated fromthe overall spectra by dividing the total power in each band bythe bandwidth in Hertz. The results are expressed as a ratio ofthe MSE in Band I (4–8 Hz) and Band II (8–12 Hz). Statistical signif-icance of MSE values was calculated using a Wilcoxon–Mann–Whitney U test, given the non-Gaussian distribution of the data.

2.4. MEG-independent component analysis

The power spectrum provides information about the dominantfrequencies of the signals that arise from the interaction of manyneuronal circuits acting as separate independent sources. To iden-tify and characterize the spectral characteristics of these indepen-dent sources, we used a technique known as independentcomponent analysis (ICA). ICA is a mathematic methodology thatcan disentangle each independent component from the rest, basedon their particular electrical signature [4]. Such analysis, when ap-plied to the MEG recordings, generates a set of electrical patterngroups or independent components (ICs). Other approaches, suchas the beamformer technique in which a frequency band chosenas a first step, were not used here, as ICA is a more thorough ap-proach to evaluate the complete physiological range of frequencies.ICA was implemented on the MEG recordings using the InfomaxEEGLab algorithm [46] as described in Schulman et al. [70]. Foreach recording, this method generated a set of ICs that are rankedaccording to their projected variance, which reflects their contribu-tion to the overall signal. The component with the lowest variance,IC01, contributes the most to the total signal projected to the MEGsensors. The contribution for IC01 was 5.0 ± 1.0% (n = 7) for the pa-tients and 9.5 ± 1.17% (n = 7) for the controls (p < 0.05, student’s t-test) that is a rather substantial signal level. Each of these ICs canthen be localized anatomically.



K.D. Walton et al. / PAIN�

xxx (2010) xxx–xxx 3

ARTICLE IN PRESS

2.5. Localization of independent components

ICs meeting the following criteria were localized for each sub-ject: (1) the frequency of the IC corresponded to a dominant peakin the power spectrum (see Fig. 2 for examples), (2) the sensor dis-tribution of the IC had a dipolar magnetic configuration, and (3) theIC had a low projected variance (rank <40). The contribution to theoverall projection of IC01–IC40 was 48 ± 1.8% (n = 14). Each com-ponent was localized onto a population-average T1 MRI using theinformation from the fiducial markers and head shape as describedin Schulman et al. [70]. Briefly, a volumetric source space was con-structed from a population-averaged MRI [48]. This probabilisticsource space was transformed onto a subject-specific coordinatesystem (using the head shape data) and scaled to match the dis-tances between the nasion and two periauricular points of eachindividual subject (using the fiducial marker information). Theleadfield matrix was calculated for each experiment using a spher-ical volume conductor model and inverse solutions were calculatedfor components of interest [19,61]. For visualization, the resultswere smoothed with a three-dimensional Gaussian kernel and lin-early interpolated on the reconstructed gray-matter surface of ahigh resolution population-averaged MRI previously segmentedinto gray and white matters with the FreeSurfer software package[12]. The results were projected onto partially inflated brains.When activity was localized to a sulcus, activity was also projectedonto a completely inflated brain (as in Fig. 3B). Matlab (The Math-works, Natick MA, USA) and custom in-house software were usedfor all analyses. JMP (SAS Institute, Cary NC, USA) was used forthe statistical analysis, all results are given as mean ± SEM, Stu-dent’s t-test was used to determine statistical significance.

2.6. Selection of independent components

An individual unfamiliar with the actual peripheral localizationor nature of the patient’s pain examined the localized ICs. Two setsof questions were addressed. The first concerned an unbiaseddetermination of localization, that is, (1) Are any ICs localized toprimary somatosensory cortex (S1)? Following its localizationtwo further questions were asked: (2) Does this activity corre-spond to the cortical region corresponding to the reported painsite? (3) What was the frequency of the somatosensory activity?

Table 2Global power spectrum and selected independent component (IC) data for patients.

Pt MEG Power spectra peak Somatosensory ICs (Fig. 3) Delta

<4 Hz 4–8 Hz 9–12 HzEO/EC

Freq.(Hz)

Region Freq.(Hz)

P15 4D No Yes –P16 4D No Yes –P24 4D No Yes 0.73P25 4D Yes No 0.75P29 4D No Yes 0.89

P30 4D No No 0.56P34 CTF Yes Yes 0.29 7.6, 16 Right S1, M1, SA 2–5,P35 CTF Yes Yes 0.54 7.9, 15.8 Rt S1, M1 3, 15

P37 CTF Yes Yes 0.18 10.5(7–12)

Bilateral S1 2.6, 8

P38 CTF Yes Yes 0.71 <4, 12.3 Rt S1, M1 2.3, 7P44 CTF No Yes 1.1 <4, 8.2 Bilateral S1, M1 2.3–9

P45 CTF Yes Yes 0.56 7.6 Lf S1, M1 2.3–8P47 CTF Yes Yes 0.73 <4, 8.7 Bilateral S1, SA Sup

parietal<4, 9

9–12 Hz EO/EC, ratio of mean power spectra at 9–12 Hz with eyes open (EO) to eyes csomatosensory associate cortex; Sup, superior; TP, temporal pole. Peaks calculated for 2


�(2010), doi:10.1016/j.pain.2010.02.023

The second set of questions addressed dominant frequency andwas independent of location. They concerned two frequencyranges (delta/theta) and (alpha). Concerning activity in the delta/theta range, they were (1) Are any ICs characterized by activityin the delta, <4 Hz, and/or theta, 4–9 Hz, frequency ranges? (2) Ifpresent, where is the low frequency activity localized?

Concerning activity in the alpha frequency range they were (1)Are any ICs dominated by alpha range activity (9–13 Hz)? (2)Where is this activity localized? Activity in the alpha range is thedominant activity in awake, healthy adults [56]. This allowed usto find if such normal activity was present in CRPS I patients andif its localization was similar to that of healthy controls.

This approach generated three IC images for each of the 7 pa-tients and included one somatosensory IC, one low frequency IC,and one alpha frequency IC. If more than one IC had a similar local-ization or dominant frequency, the IC with the lowest variance wasselected.

3. Results

3.1. Clinical description

The demographics and clinical description of the patients aresummarized in Table 1. CRPS I was related to the upper extremityin 3 patients (P25, P38, and P45) and to the lower extremity in 8patients. In 2 patients (P35 and P37) the disease had spread to bothlower extremities. At the time of the MEG recording all the patientshad significant pain; complaints measuring >6 on the unidimen-sional 100-mm visual analog scale (VAS). The duration of the dis-ease spanned from 3 to 120 months, with a mean of25.8 ± 8.9 months. All the patients were taking one or several ofthe following medications at the time of the MEG recording:anti-convulsants (12), anti-depressants (9), a non-steroidal anti-inflammatory (NSAID) (8), opioids (7), and/or muscle relaxants(3) (Table 1).

The summary data obtained from MEG recordings for all the pa-tients are given in Table 2. Table 2 also includes which MEG instru-ment was used for the recording. Basic MEG data analysis wasmade on all patients. Given that the resolution of the data recordedwith the CTF instrument was better than that recorded with the 4Dinstrument (due to the increased number of channels and higher

/theta ICs (Fig. 4) Alpha ICs (Fig. 5)

Region Freq.(Hz)

Region

7 Bilateral OFCT 10.4 Right occipitalBilateral OFTC 10.8 Bilateral mesial occipital,

Sup parietal.5 Bilateral OFTC, Rt TP 10.3 Bilateral mesial occipital

, 13 Bilateral OFC, Rt TP 9.9 Rt occipital.4 Bilateral OFC. Lf TP 9.8 Bilateral mesial occipital,

Sup parietal.2 Bilateral OFTC; frontal pole 9.1 Lf occipital, Rt Sup parietal

Bilateral OFTC and Sup parietal 9.1 Bilateral occipital, Sup parietal

losed (EC); Inf, inferior; OFTC, orbitofrontal–temporal cortex; Rt, right; Lf, left; SA,.5–20 Hz using algorithm (Igor Pro).





ARTICLE IN PRESS

signal to noise ratio), the data recorded with the CTF instrumentwere selected for inclusion in the figures.

3.2. Power spectra

We found that abnormal brain rhythms were present in CRPSpatients. This was determined by comparing the global powerspectra for the patients and controls. The individual power spectrafrom 7 controls with their eyes closed are superimposed in Fig. 1A.Note that the dominant peak is in the alpha frequency range(9–13 Hz) as normally seen in healthy adults [56]. Individualpower spectra for 7 CTF patients are superimposed in Fig. 1B. Notethat (a) the dominant peak is at a lower frequency than for the con-trols, (b) peaks are present at low frequencies (<5 Hz), and (c) thereis more variability among the patients than the controls. The differ-ence between the two groups is clearly seen in Fig. 1C where thegroup mean power spectra are superimposed. The mean MSE ratio(Fig. 1D) was larger in the CRPS patients than in the controls. Com-parisons using a non-parametric Mann–Whitney U test indicatedthat the differences in MSE ratios between the two groups werestatistically significant (*p < 0.009).

The power spectra shown in Fig. 1B are shown individually inFig. 2 for EC (red traces) and EO (blue traces) recordings. Note thatin the EC recordings there are low frequency peaks (�) as well as aclear alpha peak (d). In one case (P44, Fig. 2D) there were severaltheta range peaks (one is marked) and in another (P47, Fig. 2F) thelow frequency peak was on the rising phase of the alpha peak.Thus, there was abnormal low frequency rhythmic activity present

Fig. 1. Spectral analysis of spontaneous activity for patients and controls. (A) SuperpositiMEG sensors as a function of frequency; (units, 10 � log fT/sqrt(Hz)) recorded from 7 copeak of brain activity is in the alpha frequency range (9–13 Hz). (B) Superposition of the mcontrols. Note that in addition to the alpha peak, there is activity at lower frequencies an(red) and controls (black). This shows that the brain activity is shifted to a lower frequenc(4–8 Hz) vs. Band II (8–12 Hz) was larger in the CRPS patients (0.72 ± 0.12, n = 12) than


�(2010), doi:10.1016/j.pain.2010.02.023

in the brains of these patients while they were sitting quietly. Typ-ically, the main difference between the EC and EO power spectrumin healthy adults is a reduction in the dominant alpha peak ampli-tude in the EO condition [56]. This reduction, expressed as theEO:EC power ratio at 9–12 Hz (Table 2), was seen in all but one pa-tient (P44, Fig. 2D and Table 2). There were two notable differencesbetween the EC and EO power spectra in patients: (1) the low fre-quency peaks were not sensitive to whether the recordings weremade with the EC or the EO (Fig. 2), and (2) in the absence of deltapeaks, the power spectra at low frequencies (<5 Hz) increased inthe EO condition (Fig. 2A, D–F, blue arrowheads).

3.3. Independent components and their localization

To better understand these findings in terms of neuronal func-tion, we asked which brain areas were generating this rhythmicactivity. Toward this end, we implemented ICA localization of theMEG data recorded with the eyes closed. Of the ICs generated foreach recording fulfilling our criteria, we selected three ICs as givenin Section 2.6. First, we show the data for one patient in detail(Fig. 3). This is followed by the localization of the selected somato-sensory ICs (Fig. 4), low frequency ICs (Fig. 5), and alpha frequencyICs (Fig. 6).

As an illustrative example of the procedures followed, three se-lected ICs for one patient (P34) and the projection of each IC onto astandard semi-inflated brain are shown in Fig. 3. Since one of thecriteria for selecting ICs for detailed analysis was their correspon-dence to major power spectra peaks, these ICs are superimposed

on of power spectra mean power of spontaneous brain activity recorded from all thentrols when their eyes were closed using the CTF MEG instrument. Note the majorean power spectra recorded from 7 CRPS I patients under the same conditions as the

d there is more variability in the spectra. (C) Plot of mean power spectra for patientsy in the patients compared to the controls. (D) Mean spectral energy ratio for Band Iin the controls (0.40 ± 0.06, n = 13) (p < 0.009 Wilcoxon–Mann–Whitney U test).



Fig. 2. Spectral analysis of patients. The power spectra for the recording with the eyes closed (EC, red) and the eyes open (EO, blue) are shown for 7 patients (units, 10 � log fT/sqrt(Hz)). In addition to the normal activity in the alpha frequency range (d), there is clear, abnormal activity at lower frequencies in all patients. In five cases clear peaks ofactivity are present (A–E, G, �). In one case, (F) the peak is on the rising phase of the alpha peak. When the recordings were made with the EC (blue traces), the alpha peakdecreased in all but one case (D, P44). Note, however, that the low frequency peaks were not as affected and for those without delta peaks, activity in the delta range (<4 Hz)increased in the EC recordings (A, E, F, arrowheads).



ARTICLE IN PRESS

on the power spectrum (Fig. 3A, C, E; ICs black traces, power spec-trum, red traces).

The selected somatosensory IC, IC01, corresponds most stronglyto the theta peak in the power spectrum (Fig. 3A, �). There is also


�(2010), doi:10.1016/j.pain.2010.02.023

activity at 4–5 Hz (j) and increased power in the higher, beta fre-quency range (13–20 Hz, j). As shown in Fig. 3B, this activity waslocalized to right S1 within the central sulcus (CS) and post-centralsulcus (PCS) and posterior to this in somatosensory association



Fig. 3. Three selected independent components (ICs) and their localization for a patient with pain in her left foot and ankle (P34). IC localization is projected onto a partiallyinflated standardized brain. The brain surface that best illustrates the localization of the selected IC is shown. When the activity is in a sulcus, an inset shows the localizationin a fully inflated brain (gray areas are sulci). (A, C, and E) Superposition of the power spectrum (red; units 10 � log fT/sqrt(Hz)) and IC (black; arbitrary units). (A) Selectedsomatosensory IC (IC01) with a peak near 7 Hz (�) and lower levels of activity in delta (<4 Hz, j) and beta (13–20 Hz, j) ranges. (B) Localization of IC01 to right S1 and M1along CS and PCS extending to the mesial surface and over the right SA in the superior parietal cortex (see expanded views). (C) Selected low frequency IC (IC13) with increasedlow frequency activity (<5 Hz) and peak in theta range (�). (D) Localization of IC13 to OCF bilaterally and left TP. (E) Selected alpha IC (IC03) with peak corresponding to themajor peak in the power spectrum. (E) Localization of IC03 to right occipital cortex. Abb. CS, central sulcus; L, left; M1, primary motor cortex; OFC, orbital frontal cortex; R,right; S1, primary somatosensory cortex; SA, association somatosensory sensory cortex; TP, temporal pole.



ARTICLE IN PRESS

cortex (SA). Within the CS there is also activity in primary motorcortex (M1) as shown in the dorsal and mesial brain views andthe inserts showing the localizations on the fully inflated brain.These localizations corresponded well with the left foot and anklepain in this patient (Table 1).

The selected delta/theta component, IC13, has two elements(Fig. 3C, black trace), corresponding to the theta peak (�) and in-creased delta power (<5 Hz) in the absence of peaks. As shown inFig. 3D, this low frequency activity was localized to ventromedialprefrontal cortex (VMPFC), specifically the mesial and ventral orbito-frontal cortex (OFC) bilaterally and to the left temporal pole (TP).

The selected alpha component, IC03, has a major peak at 10 Hz(Fig. 3E, d) with no low frequency activity. Its shape is very similarto the major component of the power spectrum. This 10 Hz activitywas localized to the right occipital cortex as seen on the lateral anddorsal views of the brain (Fig. 3F).


�(2010), doi:10.1016/j.pain.2010.02.023

3.4. Somatosensory-independent components

IC localized to S1 is shown in Fig. 4 for the other patients. Activ-ity was localized to the central sulcus (CS, Fig. 4B–E) and includedprimary motor cortex (M1) in some cases (Fig. 4A–D). Such activitywas unilateral in four cases (Fig. 4A–C) and bilateral in three cases(Fig. 4D–F). In P47 (Fig. 4F) there was, in addition, activity in SA andsuperior parietal cortex bilaterally. These localizations correlatedwell with the region of pain reported by the patients (Table 1).

Considering the frequency profile (lower part of each panel), inthree cases the dominant peak of rhythmic activity was in the the-ta range (Fig. 4A, B, and D). In those cases with a dominant alpharange peak (Fig. 4C, E, and F) low frequency activity was also pres-ent in the delta (Fig. 4C and F) or theta (Fig. 4E) range. The locali-zation and frequency of the major elements of each IC are given inTable 2.



Fig. 4. Somatosensory-independent components (ICs) and their projected localization on a partially inflated standardized brain. When the activity is in a sulcus, an insetshows the localization in a fully inflated brain (gray areas are sulci). The IC is shown at the bottom of each panel. (A) Localization of IC in a woman with right forearm andwrist pain. Activity largely in left S1 and M1 extending into right premotor cortex is shown in dorsal and posterior views. (B) Localization of IC in a woman with left kneeinjury and bilateral leg pain. This activity, in the theta, alpha, and beta frequency ranges, was localized over right S1, M1, and right mesial supplemental motor cortex (shownin mid-sagittal view). (C) Localization of an IC in a woman with pain on the left side of her body. This IC, with activity in the delta, theta and alpha ranges, was localized toright S1 and M1 as shown in dorsal and right lateral views. (D) Localization of an IC in a woman with pain on the dorsal aspect of her left foot. This IC, with delta, theta andalpha range activity, was localized to mesial S1 bilaterally and extended across the right CS (arrowheads) to M1. (E) Localization of an IC in a woman with bilateral pain in herankles and feet. This IC, with a broad peak spanning theta and alpha ranges, was localized to mesial S1 and mesial occipital cortex bilaterally. (F) Localization of an IC in awoman with pain in her right knee. This activity, in the delta and theta ranges, was localized bilaterally to S1, SA and superior parietal cortex. Abbreviations as in Fig. 3.



ARTICLE IN PRESS

3.5. Low frequency-independent components

ICs with dominant activity in the delta and theta frequencyranges were found in every patient as shown in Fig. 5 (andFig. 3B for P34). In every case, this activity was localized to theorbitofrontal cortex (OFC, labeled in panel B) and the temporal pole(TP, labeled in panels A and B). In one patient (P47) activity alsoprojected bilaterally to SA cortices in the region of the marginalbranch of the cingulate sulcus (Fig. 5F, CingS).

The ICs are shown in the lower right part of each panel in Fig. 5.There is a large delta range component and theta range activity ineach case. There is a large alpha component in P47 (Fig. 5F). Thelocalization and frequency of the major elements of each compo-nent is given in Table 2.


�(2010), doi:10.1016/j.pain.2010.02.023

3.6. Alpha frequency-independent components

ICs with dominant activity in the alpha frequency range werefound in all the patients and controls. For each patient, the compo-nent is shown under its projected localization (Figs. 6 and 3C). Thepeak in each case was near 10 Hz (Table 2). The alpha activity waslocalized to the right or left occipital cortex in three patients (Figs.3F and 6A and C) and bilaterally on the mid-sagittal plane in oth-ers. In P44 and P47 (Fig. 6D and F) the activity extended acrossthe parieto-occipital sulcus (OPS, marked in A and B) into the SAin the medial or superior parietal cortex. Thus, as typically seenin normal adults [56], alpha range activity in these patients waslocalized to posterior regions of the brain. The localization and fre-quency of the selected alpha peak components are given in Table 2.



Fig. 5. Low frequency-independent components (ICs) and their projected localization. Each panel shows a mid-sagittal view of the partially inflated standardized brain andthe ventral view including the OFC and TP. Each IC is shown in the lower right of each panel. (A–F) Note that all patients have activity in the mesial OFC bilaterally thatextends to the medial aspect of the ventral OFC. TP activity is on the right side in 3 patients (B, C, and E), on the left side in one patient (A), and bilateral in one patient (F). InP47 (F) there is also activity bilaterally in the sensory association cortices in the region of the marginal branch of the cingulate sulcus (CingS). The IC includes activity in thedelta and theta frequency ranges. Alpha range activity is also present in one patient (F, P47). Abbreviations as in Fig. 3.



ARTICLE IN PRESS

4. Discussion

The main common thread in this study concerns the presence ofspontaneous abnormal low frequency rhythmic magneto-encepha-lographic activity in CRPS I patients. They all had significant intrac-table chronic neuropathic pain (NPP) that was resistant to classicneuropathic medications and all reported pain during the MEGrecording sessions. MEG recordings showed low frequency rhyth-mic activity localized to somatosensory and limbic brain regions.This is helpful in providing an objective common denominatorfor the diagnosis of NPP in CRPS I patients. The dynamics of thisabnormal rhythmic activity sheds light on the neuronal basis forsuch altered CNS function.

4.1. Spontaneous rhythmic activity

MEG power spectra, reflecting activity in the entire brain, werecharacterized by the presence of well-defined peaks or increasedpower in the delta, theta, and alpha frequency ranges (Figs. 1and 2). The abnormal increased low frequency oscillations occurred


�(2010), doi:10.1016/j.pain.2010.02.023

over a wide range (�3–9 Hz, Fig. 1C). As is typical of healthy adults[56], the alpha frequency peak was distributed to the occipital andparietal cortices (Figs. 3F and 6) and decreased in power with theEO in all but one patient (Fig. 2 and Table 2). In contrast, the peak-to-peak amplitude of low frequency activity was similar for EOand EC (Fig. 2) and activity at the lowest frequencies increased withthe EO (Fig. 2). These findings indicate that the neuronal mecha-nisms underlying the low frequency rhythmic activity are function-ally different from those underlying normal alpha activity.

4.2. Somatosensory low frequency rhythmic activity

Our finding of somatosensory activity corresponding to the re-ported region of spontaneous pain is consistent with such localiza-tion in evoked pain studies (see [71]). However, domination by lowfrequency activity (Figs. 3A and 4) suggests that these neurons donot participate directly in pain localization. Rather, these neuronsinduce increased activity in adjacent cortical regions through anedge effect [37,39]. A reduction in the normal lateral inhibitionwould force adjacent cortical areas into spontaneous and pro-



Fig. 6. Alpha frequency range-independent components (ICs) and their localization. Each panel shows the view of the partially inflated standardized brain that best illustratesthe localization of the selected IC. (A) Localization of IC to left occipital and right superior parietal cortices. (B) Localization of IC to the mesial parietal cortex bilaterally. (C)Localization of IC to right occipital cortex. (D) Localization of IC to bilateral mesial parietal cortex extending into occipital cortex. (E) Localization of IC to mesial occipitalcortex. (F) Localization of IC to dorsal–mesial occipital and superior parietal cortex. Abb. OPS, occipital–parietal cortex and as in Fig. 3.



ARTICLE IN PRESS

tracted high frequency oscillations resulting in a constantly pres-ent sensation. This interpretation is consistent with MEG [30,43]and EEG [60] studies of patients with unilateral CRPS I showingthat the CNS signal evoked by sensory stimulation of the affectedlimb is greater than that evoked by simulation of the unaffectedlimb. Changes of S1 are also seen in other types of NPP[13,14,45,74]. Pain localization is provided by such somatosensoryactivation while the emotional component is provided by activityin limbic regions.

4.3. Limbic system low frequency rhythmic activity

Indeed, a significant finding was ICs with clear theta and deltaactivity that was independent of pain localization. Such activitywas consistently localized over the OFC and region of the TP (Figs.3B and 5). Projection of activity to the TP may reflect activation ofthe nearby amygdala to which the OFC is connected [32]. Localiza-tion to OFC has been encountered in studies of acute pain and wasinterpreted as being related to the affective component of pain[78]. An fMRI study of neutral, pleasant, or painful touch found thatwhile neutral touch was most effective in somatosensory cortexactivation, painful or pleasant touch was most effective in OFC acti-vation [64]. One brain area incorporating the OFC, the VMPFC, hasbeen linked to the emotional aspects of pain as seen in an emo-tional decision-making task [3]. Also, an MRI study in CRPS pa-tients found gray matter atrophy in a region that included theright VMPFC [17].


�(2010), doi:10.1016/j.pain.2010.02.023

The above suggests a role for the VMPFC in NPP. The OFC itselfreceives input from (and projects to) the mediodorsal thalamic nu-cleus [15,29], and components of the pain matrix [32]. Low fre-quency activity in mediodorsal thalamus would lead to lowfrequency rhythmic activity in OFC. While the manner in whichthis connectivity ultimately affects the emotional color of thesomatosensory sensation is far from clear, the present finding, ofconsistent activation of the medial OFC, may reflect input fromthe limbic system, and the affective aspect of the pain. A recentmeta-analysis of imaging data suggests that medial frontal areas,as seen in this study, are more closely associated with limbic acti-vation than are lateral frontal areas [31]. The location of the lowfrequency oscillations to components of the thalamocortical painmatrix correlates well with the recent meta-analyses of functionalimaging studies [2,54]. Also, a reduction of spontaneous pain withtreatment was correlated with a reduced fMRI signal in brain areasconcerned with emotion and reward, including the OFC [16].

4.4. Possible role of medications

We considered the possibility that increased low frequencyactivity might be attributable to the medications taken by thepatients (Table 1). A study of anti-convulsant drugs in epilepsypatients and healthy controls reported in an increase in relativeEEG power in the delta and theta ranges in patients at the onsetof drug delivery and following drug withdrawal [66]. However,they found no significant difference between healthy controls





ARTICLE IN PRESS

and patients with long-term, stable monotherapy [66]. Our pa-tients fit this last category. The other drug group known to affectbrain activity is anti-depressants. Spectral power in the low andalpha frequency ranges can either increase or decrease, depend-ing on the drug and time after administration [65]. Indeed, whileunder some conditions medication may generate false positives,in all cases studied the abnormal low frequency activity was re-lated to the specific anatomical loci in accordance with the re-ported pain (Fig. 4). Thus, we conclude that the present MEGfindings are related to the NPP pathophysiology.

4.5. CRPS I as a disorder of thalamocortical interplay

An attractive interpretation of these data is that CRPS I re-flects a neurological ‘‘disconnection syndrome” resulting in adisturbance in thalamocortical interplay and a chronic dynami-cally recurrent TCD. Electrophysiologically, TCD is characterizedby hyperpolarization of thalamic neurons, low frequency reso-nant recurrent interaction between thalamic and cortical neurons[35], and the presence of an edge effect. Thalamic neuronalhyperpolarization occurs by excess inhibition [34], disfacilitationfrom thalamic deafferentation [35,76], or block of excitatory li-gand-gated channels [72]. Low frequency rhythmic MEG activityas seen here is, we propose, the result of a functional channelop-athy. Thus, a chronic deinactivation of thalamic Cav3.1 channels[7,40] results in low frequency spontaneous thalamic rhythmic-ity. Indeed, low frequency thalamic bursts have been recordedfrom NPP patients [26].

Thalamic neuron hyperpolarization is consistent with the find-ing that somatosensory cortical neurons in CRPS I patients have areduced sensitivity to tactile perception [58] and a reduced corticalrepresentation of the affected body region when evoked by naturalor electrical stimulation [30,43,58,60]. Thus, this reduction may re-sult not only from an impoverished input but also from a reducedresponsiveness of the remaining cells to the small natural inputsthey might receive.

There is an emerging theme of TCD in NPP. Low frequencyoscillations and shift of the power spectra toward lower frequen-cies have consistently been found in other NPP states as recordedfrom cortex [6,10,69,70,73] and thalamus [20,25–27,33,69]. Astudy of patients with chronic, severe NPP reported a high tem-poral coherence between central lateral thalamic field potentialsand cortical EEG oscillations at 4–9 Hz [68]. Such high thalamo-cortical coherence supports the role of the thalamus in synchro-nizing the cortical oscillations seen in animal studies [40].

An edge effect has been observed in other types of NPP as in-creased beta activity [28,69,73] and as increased gamma activityin tinnitus [9,39,63,77] and migraineurs [8]. This edge effect hasbeen proposed to generate the positive symptoms in pain and allo-dynia [27,36,70], in movement [53,67], and in psychiatric disorders[36,38,39,77]. Thus, a common neuronal mechanism may underliethese disorders. TCD in various disorders is distinguished by the re-gion of thalamic nucleus with low frequency oscillations [26]. Thepresence of high frequency activity is a subject of interest for fu-ture MEG studies of CRPS I.

4.6. Clinical implications and conclusions

The finding of brain dysfunction in CRPS I has a diagnostic valueproviding an objective, biological marker of the disease and may sig-nificantly contribute to the development of new lines of treatmentbased on the neuronal mechanisms underlying TCD. Pharmacologi-cally, since T-type calcium channel activation underlies the low fre-quency oscillations, T-type calcium channel blockers are promisingtargets for the development of effective analgesics [22].


�(2010), doi:10.1016/j.pain.2010.02.023

Concerning stimulation, any stimulus resulting in thalamic neu-ron depolarization, and thus normal rhythmicity and responsive-ness, will be beneficial in CRPS I patients. The effects of spinalcord and brain stimulation in phantom pain [62] probably resultfrom thalamic depolarization by collateral afferent activity andpossible sprouting-type plasticity. Also based on TCD circuitry, atherapeutic microlesion of the Field of Forel may provide lastingpain relief in CRPS I as in other types of NPP [27,28]. Followingcampectomy, abnormal low and high frequency activity and re-ported pain level gradually decrease together [68,69].

The findings reported here are consistent with those in otherchronic neuropathic pain conditions [6,10,28,68,70,73] and furtherour understanding of a central mechanism of NPP.

Conflict of interest

The authors do not have any conflicts of interest.

Acknowledgments

This work was supported by NYU Medical Center General Clin-ical Research Center (NCRR RR00096) and NIH Grant NS13742. Wethank Drs. D. Rojas-Soto and D. Levacic for valuable assistance.

References

[1] Apkarian AV, Baliki MN, Geha PY. Towards a theory of chronic pain. ProgNeurobiol 2009;87:81–97.

[2] Apkarian AV, Bushnell MC, Treede RD, Zubieta JK. Human brain mechanisms ofpain perception and regulation in health and disease. Eur J Pain2005;9:463–84.

[3] Apkarian AV, Sosa Y, Krauss BR, Thomas PS, Fredrickson BE, Levy RE, HardenRN, Chialvo DR. Chronic pain patients are impaired on an emotional decision-making task. Pain 2004;108:129–36.

[4] Bell AJ, Sejnowski TJ. An information–maximization approach to blindseparation and blind deconvolution. Neural Comput 1995;7:1129–59.

[5] Birklein F, Handwerker HO. Complex regional pain syndrome: how to resolvethe complexity? Pain 2001;94:1–6 [see comment].

[6] Boord P, Siddall PJ, Tran Y, Herbert D, Middleton J, Craig A.Electroencephalographic slowing and reduced reactivity in neuropathic painfollowing spinal cord injury. Spinal Cord 2008;46:118–23.

[7] Chemin J, Monteil A, Perez-Reyes E, Bourinet E, Nargeot J, Lory P. Specificcontribution of human T-type calcium channel isotypes (alpha(1G), alpha(1H)and alpha(1I)) to neuronal excitability. J Physiol 2002;540:3–14.

[8] Coppola G, Ambrosini A, Di Clemente L, Magis D, Fumal A, Gerard P, Pierelli F,Schoenen J. Interictal abnormalities of gamma band activity in visual evokedresponses in migraine: an indication of thalamocortical dysrhythmia?Cephalalgia 2007;27:1360–7.

[9] De Ridder D, De Mulder G, Verstraeten E, Seidman M, Elisevich K, Sunaert S,Kovacs S, Van der Kelen K, Van de Heyning P, Moller A. Auditory cortexstimulation for tinnitus. Acta Neurochir Suppl 2007;97:451–62.

[10] Drewes A, Gratkowski M, Sami S, Dimcevski G, Funch-Jensen P, Arendt-NielsenL. Is the pain in chronic pancreatitis of neuropathic origin? Support from EEGstudies during experimental pain. World J Gastroenterol 2008;14:4020–7.

[11] Dubois M, Rojas-Soto D, Garcia J, Levacic D, Walton K, Llinas R. Abnormal brainactivity in patients with complex regional pain syndrome (CRPS) type I.Abstract Viewer/Itinerary Planner, Program No. 3463, Washington, DC; 2008.

[12] Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, van der KouweA, Killiany R, Kennedy D, Klaveness S, Montillo A, Makris N, Rosen B, Dale AM.Whole brain segmentation: automated labeling of neuroanatomical structuresin the human brain. Neuron 2002;33:341–55.

[13] Flor H, Braun C, Elbert T, Birbaumer N. Extensive reorganization of primarysomatosensory cortex in chronic back pain patients. Neurosci Lett1997;224:5–8.

[14] Flor H, Elbert T, Knecht S, Wienbruch C, Pantev C, Birbaumer N, Larbig W, TaubE. Phantom-limb pain as a perceptual correlate of cortical reorganizationfollowing arm amputation. Nature 1995;375:482–4.

[15] Fuster J. The prefrontal cortex. New York: Raven Press; 2000.[16] Geha PY, Baliki MN, Chialvo DR, Harden RN, Paice JA, Apkarian AV. Brain

activity for spontaneous pain of postherpetic neuralgia and its modulation bylidocaine patch therapy. Pain 2007;128:88–100.

[17] Geha PY, Baliki MN, Harden RN, Bauer WR, Parrish TB, Apkarian AV. The brainin chronic CRPS pain: abnormal gray–white matter interactions in emotionaland autonomic regions. Neuron 2008;60:570–81.

[18] Ghazal T, Moran K, Walton K, Llinas R, Dubois M. Patients with CRPSdemonstrate thalamocortical dysrhythmia: a magneto-encephalographic(MEG) study. In: 23rd annual meeting. New Orleans: American Academy ofPain Medicine; 2007.





ARTICLE IN PRESS

[19] Gorodnitsky IF, George JS, Rao BD. Neuromagnetic source imaging withFOCUSS: a recursive weighted minimum norm algorithm. ElectroencephalogrClin Neurophysiol 1995;95:231–51.

[20] Gucer G, Niedermeyer E, Long DM. Thalamic EEG recordings in patients withchronic pain. J Neurol 1978;219:47–61.

[21] Harden R, Bruehl S, Stanton-Hicks M, Wilson P. Proposed newdiagnostic criteria for complex regional pain syndrome. Pain Med2007;8:326–31.

[22] Hildebrand M, Snutch P. Contribution of T-type calcium channels to thepathophysiology of pain signalling. Drug Discov Today Dis Mech2006;3:335–41.

[23] Janig W, Baron R. Complex regional pain syndrome is a disease of the centralnervous system. Clin Auton Res 2002;12:150–64.

[24] Janig W, Baron R. Experimental approach to CRPS. Pain 2004;108:3–7.[25] Jeanmonod D, Magnin M, Morel A. Thalamus and neurogenic pain:

physiological, anatomical and clinical data. Neuroreport 1993;4:475–8[erratum appears in Neuroreport 1993;4:1066].

[26] Jeanmonod D, Magnin M, Morel A. Low-threshold calcium spike bursts in thehuman thalamus. Common physiopathology for sensory, motor and limbicpositive symptoms. Brain 1996;119:363–75.

[27] Jeanmonod D, Magnin M, Morel A, Siegmund M. Surgical control of the humanthalamocortical dysrhythmia: I. Central lateral thalamotomy in neurogenicpain. Thalamus Relat Syst 2001;1:71–9.

[28] Jeanmonod D, Sarnthein J, Magnin M, Stern J, Lebzelter C, Aufenberg C. Chronicneurogenic pain: thalamocortical dysrhythmia mechanisms and their surgicaltreatment. Thalamus Relat Syst 2005;3:63–70.

[29] Jones E. The thalamus. Cambridge University Press; 2007.[30] Juottonen K, Gockel M, Silen T, Hurri H, Hari R, Forss N. Altered central

sensorimotor processing in patients with complex regional pain syndrome.Pain 2002;98:315–23.

[31] Kober H, Barrett LF, Joseph J, Bliss-Moreau E, Lindquist K, Wager TD. Functionalgrouping and cortical–subcortical interactions in emotion: a meta-analysis ofneuroimaging studies. Neuroimage 2008;42:998–1031.

[32] Kringelbach ML, Rolls ET. The functional neuroanatomy of the humanorbitofrontal cortex: evidence from neuroimaging and neuropsychology.Prog Neurobiol 2004;72:341–72.

[33] Lenz FA, Kwan HC, Dostrovsky JO, Tasker RR. Characteristics of the burstingpattern of action potentials that occurs in the thalamus of patients withcentral pain. Brain Res 1989;496:357–60.

[34] Llinas R, Jahnsen H. Electrophysiology of mammalian thalamic neuronesin vitro. Nature 1982;297:406–8.

[35] Llinas R, Ribary U, Contreras D, Pedroarena C. The neuronal basis forconsciousness. Philos Trans R Soc Lond B Biol Sci 1998;353:1841–9.

[36] Llinas R, Ribary U, Jeanmonod D, Cancro R, Kronberg E, Schulman J,Zonenshayn M, Magnin M, Morel A, Siegmund M. Thalamocorticaldysrhythmia I. Functional and imaging aspects. Thalamus Relat Syst2001;1:237–44.

[37] Llinas R, Urbano F, Leznik E, Ramizeriz R, Van Marle H. Rhythmic anddysrhythmic thalamocortical dynamics: GABA systems and the edge effect.Trends Neurosci 2005;28:325–33.

[38] Llinas RR, Choi S, Urbano FJ, Shin HS. Gamma-band deficiency and abnormalthalamocortical activity in P/Q-type channel mutant mice. Proc Natl Acad SciUSA 2007;104:17819–24 [see comment].

[39] Llinas RR, Ribary U, Jeanmonod D, Kronberg E, Mitra PP. Thalamocorticaldysrhythmia: a neurological and neuropsychiatric syndrome characterized bymagnetoencephalography. Proc Natl Acad Sci USA 1999;96:15222–7.

[40] Llinas RR, Steriade M. Bursting of thalamic neurons and states of vigilance. JNeurophysiol 2006;95:3297–308.

[41] Maihofner C, Baron R, DeCol R, Binder A, Birklein F, Deuschl G, Handwerker HO,Schattschneider J. The motor system shows adaptive changes in complexregional pain syndrome. Brain 2007;130:2671–87.

[42] Maihofner C, Forster C, Birklein F, Neundorfer B, Handwerker HO. Brainprocessing during mechanical hyperalgesia in complex regional painsyndrome: a functional MRI study. Pain 2005;114:93–103.

[43] Maihofner C, Handwerker HO, Neundorfer B, Birklein F. Patterns of corticalreorganization in complex regional pain syndrome. Neurology2003;61:1707–15.

[44] Maihofner C, Handwerker HO, Neundorfer B, Birklein F. Cortical reorganizationduring recovery from complex regional pain syndrome. Neurology2004;63:693–701.

[45] Maihofner C, Neundorfer B, Stefan H, Handwerker HO. Cortical processing ofbrush-evoked allodynia. Neuroreport 2003;14:785–9.

[46] Makeig S, Jung TP, Bell AJ, Ghahremani D, Sejnowski TJ. Blind separation ofauditory event-related brain responses into independent components. ProcNatl Acad Sci USA 1997;94:10979–84.

[47] Makela JP, Salmelin R, Kotila M, Salonen O, Laaksonen R, Hokkanen L, Hari R.Modification of neuromagnetic cortical signals by thalamic infarctions.Electroencephalogr Clin Neurophysiol 1998;106:433–43.

[48] Mazziotta J, Toga A, Evans A, Fox P, Lancaster J, Woods R. A probabilisticapproach to mapping the human brain. In: Toga AW, Mazziotta JC, editors.Brain mapping: the systems. San Diego: Academic Press; 2000.

[49] McCabe CS, Haigh RC, Ring EF, Halligan PW, Wall PD, Blake DR. A controlledpilot study of the utility of mirror visual feedback in the treatment ofcomplex regional pain syndrome (type 1). Rheumatology (Oxford) 2003;42:97–101.


�(2010), doi:10.1016/j.pain.2010.02.023

[50] McCubbin J, Vrba J, Spear P, McKenzie D, Willis R, Loewen R, Robinson SE, FifeAA. Advanced electronics for the CTF MEG system. Neurol Clin Neurophysiol2004;30:69.

[51] Merskey H, Bogduk N, editors. Classification of chronic pain. Seattle: IASPPress; 1994.

[52] Mitra PP, Pesaran B. Analysis of dynamic brain imaging data. Biophys J1999;76:691–708.

[53] Moazami-Goudarzi M, Sarnthein J, Michels L, Moukhtieva R, Jeanmonod D.Enhanced frontal low and high frequency power and synchronization in theresting EEG of parkinsonian patients. Neuroimage 2008;41:985–97.

[54] Moisset X, Bouhassira D. Brain imaging of neuropathic pain. Neuroimage2007;37:S80–8.

[55] Moseley GL, Zalucki N, Birklein F, Marinus J, van Hilten JJ, Luomajoki H.Thinking about movement hurts: the effect of motor imagery on pain andswelling in people with chronic arm pain. Arthritis Rheum 2008;59:623–31.

[56] Nunez PL, Wingeier BM, Silberstein RB. Spatial–temporal structures of humanalpha rhythms: theory, microcurrent sources, multiscale measurements, andglobal binding of local networks. Hum Brain Mapp 2001;13:125–64.

[57] Peyron R, Laurent B, Garcia-Larrea L. Functional imaging of brain responses topain. A review and meta-analysis. Neurophysiol Clin 2000;30:263–88.

[58] Pleger B, Ragert P, Schwenkreis P, Forster AF, Wilimzig C, Dinse H, Nicolas V,Maier C, Tegenthoff M. Patterns of cortical reorganization parallel impairedtactile discrimination and pain intensity in complex regional pain syndrome.Neuroimage 2006;32:503–10.

[59] Pleger B, Tegenthoff M, Ragert P, Forster AF, Dinse HR, Schwenkreis P, NicolasV, Maier C. Sensorimotor retuning [corrected] in complex regional painsyndrome parallels pain reduction. Ann Neurol 2005;57:425–9 [erratumappears in Ann Neurol 2005;57:609].

[60] Pleger B, Tegenthoff M, Schwenkreis P, Janssen F, Ragert P, Dinse HR, Volker B,Zenz M, Maier C. Mean sustained pain levels are linked to hemispherical side-to-side differences of primary somatosensory cortex in the complex regionalpain syndrome I. Exp Brain Res 2004;155:115–9.

[61] Ramirez R, Kronberg E, Ribary U, Llinas R. Recursive weighted minimum-normalgorithms for neuromagnetic source imaging using diversity measureminimization: analysis of spatial resolution. Soc Neurosci Abstr 2003:29.

[62] Ray N, Jenkinson N, Kringelbach M, Hansen P, Pereira E, Brittain J, Holland P,Holliday I, Owen S, Stein J, Aziz T. Abnormal thalamocortical dynamics may bealtered by deep brain stimulation: using magnetoencephalography to studyphantom limb pain. J Clin Neurosci 2009;16:32–6.

[63] Richter GT, Mennemeier M, Bartel T, Chelette KC, Kimbrell T, Triggs W,Dornhoffer JL. Repetitive transcranial magnetic stimulation for tinnitus: a casestudy. Laryngoscope 2006;116:1867–72.

[64] Rolls ET, O’Doherty J, Kringelbach ML, Francis S, Bowtell R, McGlone F.Representations of pleasant and painful touch in the human orbitofrontal andcingulate cortices. Cereb Cortex 2003;13:308–17.

[65] Saletu B, Gruberger J, Rajna P. Pharmaco-EEG profiles of antidepressants:pharmacodynamic studies with fluvoxamine. Br J Clin Pharmacol1983;15:369S–83S.

[66] Salinsky MC, Oken BS, Storzbach D, Dodrill CB. Assessment of CNS effects ofantiepileptic drugs by using quantitative EEG measures. Epilepsia2003;44:1042–50.

[67] Sarnthein J, Jeanmonod D. High thalamocortical theta coherence in patientswith Parkinson’s disease. J Neurosci 2007;27:124–31.

[68] Sarnthein J, Jeanmonod D. High thalamocortical theta coherence in patientswith neurogenic pain. Neuroimage 2008;39:1910–7.

[69] Sarnthein J, Stern J, Aufenberg C, Rousson V, Jeanmonod D. Increased EEGpower and slowed dominant frequency in patients with neurogenic pain. Brain2006;129:55–64.

[70] Schulman JJ, Zonenshayn M, Ramirez RR, Ribary U, Llinas R. Thalamocorticaldysrhythmia syndrome: MEG imaging of neuropathic pain. Thalamus RelatSyst 2005;3:33–9.

[71] Schwenkreis P, Maler C, Tegenthoff M. Functional imaging of central nervoussystem involvement in complex regional pain syndrome. Am J Neuroradiol2009:1–6.

[72] Serulle U, Urbano F, Lisman J, Llinas R. Increase of oscillatory inhibitorysynaptic events in thalamic relay neurons after inhibition of NMDA-mediatedtransmission. Washington, DC: Soc Neurosci; 2008.

[73] Stern J, Jeanmonod D, Sarnthein J. Persistent EEG overactivation in the corticalpain matrix of neurogenic pain patients. Neuroimage 2006;31:721–31.

[74] Tecchio F, Padua L, Aprile I, Rossini PM. Carpal tunnel syndrome modifiessensory hand cortical somatotopy: a MEG study. Hum Brain Mapp2002;17:28–36.

[75] Treede RD, Kenshalo DR, Gracely RH, Jones AK. The cortical representation ofpain. Pain 1999;79:105–11.

[76] Wang G, Thompson SM. Maladaptive homeostatic plasticity in a rodent modelof central pain syndrome: thalamic hyperexcitability after spinothalamic tractlesions. J Neurosci 2008;28:11959–69.

[77] Weisz N, Moratti S, Meinzer M, Dohrmann K, Elbert T. Tinnitus perception anddistress is related to abnormal spontaneous brain activity as measured bymagnetoencephalography. PLoS Med 2005;2:546–53.

[78] Wittig N, Kuppers R, Svensson P, Jensen T. A PET activation study ofbrush-evoked allodynia in patients with nerve injury pain. Pain2006;210:145–54.



abnormal thalamocortical activity in patients with complex...

Documents