- Email: [email protected]
Neuronal mechanisms during repetitive trigemino-nociceptive stimulation in migraine patients
Ò
PAIN 151 (2010) 97–103
www.elsevier.com/locate/pain
Neuronal mechanisms during repetitive trigemino-nociceptive stimulation in migraine patients David Aderjan 1, Anne Stankewitz 1, Arne May ⇑ Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Germany
a r t i c l e
i n f o
Article history: Received 12 February 2010 Received in revised form 10 May 2010 Accepted 19 June 2010
Keywords: Functional magnetic resonance imaging (fMRI) Migraine Habituation Trigemino-nociceptive stimulation Pain processing Endogenous pain control
a b s t r a c t Habituation deficits in various sensory modalities have been observed in migraine patients in several experimental designs. The underlying neuronal mechanisms are, however, still unknown. Past studies have used electrophysiological measures and focussed on habituation behaviour during one single session. We were interested in how repeated painful stimulation over several days is processed, perceived and modulated in migraineurs. Fifteen migraine patients and 15 healthy controls were stimulated daily with a 20 min trigeminal pain paradigm for eight consecutive days, using functional MRI performed on days one and eight and one follow-up measurement three months later. The results demonstrate that migraine patients did not differ in behavioural pain ratings compared to the controls at any time. However, functional imaging data revealed a significant difference in several brain areas over time. The activity level in the prefrontal cortex (PFC) and the rostral anterior cingulate cortex (rACC) increased in healthy control subjects from day one to day eight, whereas it decreased in migraine patients. These data suggest that several brain areas known to be involved in endogenous pain control show a completely opposite behaviour in migraine patients compared to healthy controls. These brain networks seem not to be disrupted per se in migraine patients but changed activity over time responding to repetitive nociceptive input. The alteration of pain inhibitory circuits may be the underlying mechanism responsible for the dys-functional neuronal filters of sensory input. Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Migraine is considered to be a genetically determined neurological disorder [32]. The characteristics, frequency and intensity of migraine attacks are, however, strongly affected by several endogenous and environmentally related factors [26]. Although acute headache is the predominant symptom in most cases, several other manifestations in the central nervous system such as photophobia, phonophobia, and nausea add to its complexity [20]. The pathogen mechanisms underlying migraine have been of intense research interest in the last century. There is strong evidence that several brain structures are fundamentally involved, including the trigemino-vascular system [8,16], the brainstem [4,17,30,31], and the cerebral cortex [10]. The role of the cortical excitability level in migraine disease has been studied extensively in a number of electrophysiological studies, using event-related potentials [10,25]. Most of these studies ⇑ Corresponding author. Address: Department of Systems Neuroscience, University of Hamburg-Eppendorf (UKE), Martinistrasse 52, D-20246 Hamburg, Germany. Tel.: +49 (0)40 74105 9189; fax: +49 (0)40 74105 9955. E-mail address: [email protected] (A. May). 1 These authors contributed equally to this work.
revealed that migraine patients suffer from a lack of habituation in various sensory modalities during their interictal migraine interval, whereas healthy control subjects habituated during the same period. Interestingly, in migraineurs amplitudes normalize just before and during acute headache attacks [2]. The underlying neuronal mechanisms are, however, still unclear and controversial [3,10]. A lack of habituation in migraineurs represents an altered cortical information-process, which has been hypothesized to be a consequence of the dys-functional neuronal inhibitory circuits [10]. Although habituation behaviour in migraine patients has been extensively explored in the last century, this has been done almost exclusively during one stimulation session, whereas little is known about the habituation behaviour in migraineurs over time. Most importantly, so far, no functional imaging data are available to shed light on the underlying neuronal mechanisms. This is different for healthy volunteers, where the physiological response to repetitive nociceptive input has been more fully investigated [6,7,23,28]. On the basis of these findings, we aimed to investigate migraine patients during daily trigemino-nociceptive stimulation on eight consecutive days and a follow-up measurement three months later. Behavioural pain ratings and BOLD (blood oxygen level dependence) signal changes during trigeminal pain processing
0304-3959/$36.00 Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2010.06.024
Ò
98
D. Aderjan et al. / PAIN 151 (2010) 97–103
were compared between migraineurs and healthy controls. Given the known genetically determined susceptibility of migraine patients, we hypothesized that these groups would differ in behavioural pain ratings over time and that the differences would be seen in the brain responses.
experiment. The subjects were instructed to breathe orally during the experiment to avoid fluctuations in stimulus concentrations due to respiratory airflow.
2. Materials and methods
The study phase consisted of an eight days program of daily trigemino-nociceptive stimulation and a follow-up measurement after 90 days. On day one, eight, and 90, the individuals were exposed to 15 trigemino-nociceptive stimuli (ammonia) and 15 non-noxious control stimuli (air puffs) during the event-related fMRI (Fig. 1a). The stimulation equipment and design used in the present study is described in detail elsewhere [27]. On days two to seven the stimulation was performed outside the scanner. The number of stimuli were increased to 25 for each condition (trigemino-nociceptive and controls stimuli). During these behavioural measurements which were performed in a separate room next to the scanner only behavioural intensity rating data were recorded. For intranasal stimulation an olfactometer was used which was validated in a previous study by our group [27]. This approach allows us to achieve a tolerable but effective trigeminal stimulation in the fMRI. Gaseous ammonia was applied intranasally through a thin tube and provoked a nociceptive perception in the receptive field of the first and second trigeminal nerve division. The experimental design and stimulation approach were described in detail elsewhere [27]. The software presentation (http://www.neurolabs.com, version 13.0) was used for controlling open states of specific magnet valves of the olfactometer and for programming the visual stimulation which was presented on a computer screen during the experiment. The stimulation procedure (Fig. 1b) consisted of the following events: Prior to each stimulus an attention task was implemented: After a jittered time period of 8–10 s a red fixation cross changed its colour to yellow. Participants were instructed to press a button on a button-box as soon as possible after this colour-change. Following another 8–10 s one of the two stimulus conditions (ammonia or air puff) was applied to the subjects for a period of 800 ms by opening the specific valves. The tube .was then neutralized by a following constant airflow. After a jittered time delay of 6–8 s individuals rated the intensity of perceived pain on a numerical rating scale (NRS) by operating a button-box with three fingers of the right hand. The NRS was presented as a vertical line of numbers from zero to ten (zero = no sensation; ten = unbearable pain). During the sessions outside the scanner (day 2–7) participants were instructed to verbalize ratings which were then logged in by the experimenter.
2.1. Subjects We scanned forty-five volunteers who were recruited locally. Participants gave written informed consent in accordance with the declaration of Helsinki and approved by the local Ethics Committee. Fifteen data sets had to be excluded due to pre-defined exclusion criteria regarding: clinical features (n = 3 patient has more than six attacks per month), not sufficient pain ratings (controls n = 1 and patients n = 2), technical problems with the scanner (controls n = 2 and patients n = 1), withdrawal from the experiment (patients n = 2), or movement artefacts (translation > 3 mm or rotation > 3 degrees in three axes) (controls n = 1 and patients n = 3). Complete data sets of fifteen migraineurs with and without aura and fifteen healthy control subjects were included in our statistical analyses. Demographical and clinical features are shown in Table 1. Migraine diagnosis was based on the classification of the International Headache Society [11]. None of the patients took any migraine prophylaxis. The patients were controlled for migraine attacks during the experiment, as well as for the attacks which occurred within 72 h both before and after the experiment. None of the participants had a history of internal, psychiatric or neurological disease, in particular no history of other headache or pain syndrome. We assessed any occurrence of minor pain events up to 4 weeks prior to the study period. The subjects and patients were instructed to take no medication during the study phase. The subjects were informed about the possibility to withdraw from the experiment at any time for any given reason. All the participants were remunerated for the participation. 2.2. Pre-experimental phase The participants were informed about the purpose of the study investigating the central nervous systems’ adaptive behaviour to repeated painful stimulation in the trigeminal receptive field in migraine patients. Prior to the fMRI experiment [27], the subjects were trained with this paradigm outside the scanner. Only those subjects who experienced a clearly painful sensation (which was pre-defined as a ‘‘4” or above on the NRS) following the administration of ammonia (pain condition) were included in the fMRI
2.3. Experimental protocol
2.4. Image acquisition Table 1 Demographical and clinical features and pain ratings on day one, eight and 90 are listed. Mean values of pain ratings are presented with the standard error of the mean (±sem). Migraineurs
Controls
Demography n (female) Mean age in years (±SD) Range of age
18 (9) 32.13 (9.95) 24–53
15 (8) 28.8 (7.73) 21–48
Clinical features Frequency of attacks Migraine with aura
1–6 6
Behavioural results Mean pain rating day 1 (±sem) Mean pain rating day 8 (±sem) Mean pain rating day 90 (±sem)
5.40 (0.32) 4.20 (0.34) 4.59 (0.38)
Imaging was conducted using a 3T Siemens Trio scanner with a 12-channel head coil. For functional scans, a gradient echo–echo planar imaging sequence (EPI) with TE = 30 ms and TR = 2620 ms was performed (voxel size = 3mm3, flip angle 80°, field of view 192 mm3). Each volume consisted of 42 axial slices with a thickness of 3 mm and a gap of 1 mm. High resolution T1-weighted structural images (voxel size = 1 mm3) were acquired additionally, using a MPRAGE sequence. 2.5. Analyses of behavioural data
5.96 (0.37) 4.37 (0.34) 4.37 (0.22)
Behavioural data were analysed using the Statistical Program for Social Sciences (SPSS; version 15.0; SPSS Inc., Chicago, IL, USA). The mean values and standard errors (m ± sem) of pain ratings were calculated for each measurement time point (days 1–8
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
99
Fig. 1. (a) Experimental design: trigemino-nociceptive stimulation was performed on eight consecutive days and on one follow-up three months later (day 90). Functional MRI was assessed on day one, eight and 90. (b) Shows the stimulation paradigm in detail which was performed on each experimental day. This paradigm consists of a reaction task, a stimulation exposure and a rating procedure on a numerical rating scale (NRS). After each trial, an inter-stimulus interval followed. On day one, eight and 90 the stimulation was performed during fMRI (15 stimuli were used for each of the two conditions). For behavioural measurements on day two to seven, which were recorded outside the scanner, the number of stimuli was increased to 25 as per condition.
and 90) separately for each group. Further, we compared the averaged pain ratings on days one, eight, and 90 (fMRI measurements), using a variance analysis (ANOVA) for repeated measures and tested for time and group effects as well as for an interaction between both. Three measurement points (day one, eight, and 90) were set as inner-subject factors (= time) whereas the betweensubject factor (group) split the sample into migraineurs (n = 15) and healthy subjects (n = 15). We determined whether the behavioural effect was primarily based on a reduction between the first eight days or on a lowering between days eight and 90, by setting the general-linear-model contrasts in ANOVA. We further calculated a linear regression analysis testing for significant effects over the study time course from day one to day eight for each group. For analyses of behavioural data (pain ratings), the statistical threshold was set to p < 0.05. 2.6. Image processing and statistical analysis Image processing and statistical analysis were conducted with Statistical Parametric Mapping (SPM5; Welcome Department for Imaging Neuroscience, London UK) running on matlab (Mathworks). The pre-processing included slice time correction, realignment, and spatial normalization on the Montreal Neurological Institute (MNI) stereotactic map. Finally, the data were smoothed using a 10 mm3 full-width at half-maximum (FWHM) isotopic Gaussian kernel. Analyses of single-subject data were performed using a general linear model approach (GLM). The first-level analysis of each subject consisted of the following regressors: (1) ammonia (= pain), (2) air puff, (3) reaction task (onset of the fixation cross until buttonbox press after colour changing), (4) rating procedure (onset of presenting NRS until rating was logged), and (5) movement parameters (translation and rotation around x-, y-, and z-axis). There was a jittered time interval of 10 s between the reaction task and the nociceptive input. This was done to demonstrate that the volunteers stayed alert and focussed on the task. Every single event was convolved with a canonical hemodynamic response function (HRF) with the time derivative, as implemented in SPM5. After estimation of the first-level design matrix, a linear contrast of
ammonia > baseline activity was defined. Separate contrast-images for each of the MR-recording on days one, eight, and 90 were generated, resulting in three images for each of the thirty subjects. Statistical parametric maps were interpreted by referring them to the probabilistic behaviour of Gaussian random fields. In order to analyze the group effects, contrast-images were inserted into a 2 3 flexible factorial design-matrix which consisted of two groups (controls and migraineurs) and three conditions (scan time points: day one, eight, and 90). (1) BOLD signal changes during trigemino-nociceptive stimulation during the first fMRI session in migraineurs and controls: To assess shared networks of pain-related activation in both the groups we calculated a conjunction analysis [19], including contrast-images for the ammonia > baseline activity of the first measurement point (day one) for each individual (n = 15 migraineurs and n = 15 controls). (2) Group comparison regarding BOLD signal changes over scan time points: We then performed an interaction analysis (flexible factorial design) between the group (migraineurs and controls) and the condition (time: day one, eight, and 90) testing whether migraineurs differ from controls in trigeminal pain processing over time. Contrasts weights were based on behavioural findings: decreased pain ratings were observed from day one to day eight in both groups while rating values on day 90 were comparable to those of day eight. In SPM, we tested for brain areas in which the activity increased in controls from day one to day eight and stayed constant at day 90 and decreased in migraineurs from day one to day eight and stayed constant at day 90 (see behavioural results and Fig. 3). Additionally, the opposite contrast was performed. We scanned patients at any time point during their interictal and ictal state. Due to the possible effects of acute headache on our outcome, we calculated an additional flexible 2 3-factorial design, excluding the contrast images of the patients scanned during acute head pain (n = 9). The statistical threshold for behavioural analyses were set to p < 0.05 (ANOVA for repeated measurements). The statistical threshold for the conjunction was set to p < 0.05 corrected for whole brain multiple comparisons, using familywise error (FWE) rate. For interaction analyses, we accepted an uncorrected threshold of p < 0.005 because we focussed our interest on small subcor-
Ò
100
D. Aderjan et al. / PAIN 151 (2010) 97–103
tical and brainstem structures which are highly challenging to detect [26]. 3. Results 3.1. Behavioural data Pain ratings: An ANOVA for repeated measures detected a statistically significant effect for the factor ‘‘time” (day one, eight, and 90) (F(1,13) = 15.47, p < 0.05) whereas no significant effect was revealed for the factor ‘‘group” (patients and controls) and no significant interaction effect between ‘‘group” and ‘‘time” was found. GLM (General-linear-model)-contrasts (ANOVA) revealed a significant reduction between day one and day eight, as well as between day one and day 90 (p < 0.05) (see Fig. 2). 3.2. Neuroimaging results BOLD response during trigemino-nociceptive stimulation on day one (the first stimulation session) in migraineurs and controls: A conjunction analysis revealed increased BOLD signals in several brain areas during trigemino-nociceptive stimulation, including bilateral activations in the amygdala (AMY), primary (SI) and secondary somatosensory cortex (SII), pallidum, the anterior and posterior portion of the insula cortex (IC), and in the operculum. Increased activity was further seen in the middle cingulate cortex (MCC), ipsilaterally in the cerebellum, and contralaterally in the supra-
(a)
* *
7 6 5 NRS
4 3 2
marginal and inferior frontal gyrus. MNI-coordinates of peak voxels are listed in Table 2. BOLD signal changes over time (day one, eight and 90): An interaction analysis (flexible factorial design) aiming to identify brain areas in which the activity during trigeminal pain stimulation increased in healthy controls from day one to day eight and stayed constant at day 90 and decreased in migraineurs from day one to day eight and stayed constant at day 90 revealed robust activation in the right anterior lateral prefrontal cortex (alPFC) (39, 36, 12; t(56) = 3.55), the rostral anterior cingulate cortex (rACC) (12, 48, 0; t(56) = 3.27), the midbrain at the anatomical localization of the red nucleus ( 6, 24, 6; t(56) = 3.01), and the ventral part of the upper medulla oblongata (0, 36, 51; t(56) = 2.89). Additionally, increased BOLD responses were detected in the right occipital (33, 81, 3; t(56) = 3.24) and frontal cortex (15, 51, 18; t(56) = 3.23), in the right cerebellum (24, 42, 48; t(56) = 3.38), and bilaterally in the calcarine gyrus ( 15, 51, 6; t(56) = 3.07 and 18, 48, 9; t(56) = 3.06) (p < 0.005 uncorrected) (peak voxel of activation patterns are listed in X-, Y- and Z-MNI space). The opposite contrast (areas in which activity increased in migraineurs and decreased in controls during trigeminal pain stimulation from day one to day eight and stayed constant at day 90) did not reveal any significant result (p < 0.005 uncorrected). Some patients (n = 9) reported acute headache on one of the scanning days. This is to be expected, given that the study covered 3 scanning days in a total of 90 days. We therefore used additionally a flexible factorial design which excluded these patients. This analysis did not change the results and revealed the same activation pattern as the interaction analysis including all contrast images. To exclude the possibility that patients and controls differ regarding attention, we performed an additional analysis, comparing linear contrasts modelling the activity during the reaction task at day 1 between both groups. No significant differences could be seen between migraineurs and controls (whole brain analysis; uncorrected; p < 0.005). Furthermore, reaction time scores did not explain any differences between groups and the height of the reaction time score did not correlate with the height of the BOLD signal intensity during the nociceptive stimulation neither in controls nor in migraine patients (whole brain analysis; uncorrected; p < 0.005).
1 0
4. Discussion
90
8
1
Experimental days
(b)
7 6 5
NRS
4 3 controls
2
migraineurs 1 0 1
2
3
4
5
6
7
8
Experimental days Fig. 2. The upper part of the figure (a) shows mean values of pain ratings (NRS) of healthy control subjects (black bar) and of migraine patients (white bar) on the days of scanning. A significant attenuation of pain ratings from day one to eight was seen in both groups. The pain ratings on day 90 were comparable with those on day eight. The lower part of the figure (b) shows the averaged pain ratings (NRS values) on each single day for healthy subjects (grey bar) and migraineurs (blue bar). Pain ratings showed a gradual decrease over time in both groups. Error bars indicate the standard error of the mean (sem) (see Table 1).
Contrary to our expectation that migraine patients and control subjects would differ in pain perception over time, both groups behaved the same during the eight experimental days. Migraine patients and healthy controls showed a gradual decrease of pain ratings from day one to day eight which only marginally increased again on day 90. This behaviour is identical to several studies in healthy controls using repetitive nociceptive input [6,7,23,28]. Whatever the neuronal basis for intra-session habituation may be [3,10], repeated nociceptive input over several sessions does not result in a loss of habituation in migraineurs. Of note, the neurophysiological evidences of a reduced habituation in migraine patients were issued from experiments of repetitive sensory stimuli during a single session with a very short inter-stimulus interval. By contrast, our study focussed on habituation behaviour over several days. It is therefore more than likely that the neurophysiological mechanisms behind both effects are fundamentally different. We further focussed on functional MRI data demonstrating that both groups activated a pattern of brain areas known to be involved in sensory, affective, and cognitive pain processing during the first trigeminal pain stimulation on day one [15]. Increased
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
101
Fig. 3. The interaction analysis between migraine patients and controls was contrasted to identify brain areas in which the brain activity in healthy controls during trigeminal nociceptive stimulation showed a significant increase from day one to eight (which stayed constant on day 90) whereas the activity level in these areas decreased in migraine patients. Significant BOLD signal changes were detected in the prefrontal cortex (PFC), in the rostral anterior cingulate cortex (rACC), in the midbrain (red nucleus) (NR) and in the ventral medulla. Results are plotted onto the averaged template of high resolution structural images of healthy controls and interictal migraine patients.
Table 2 MNI coordinates of the conjunction analysis of the first scanning day of migraineurs and controls. Results are FWE-corrected (p < 0.05). Anatomical structure
Coordinates of peak voxel (x, y, z in mm) Left
Prefrontal cortex (PFC) Cerebellum Rostral anterior cingulate cortex (rACC) Occipital cortex Frontal cortex Calcarine gyrus Nucleus ruber Ventral medulla (olivary nucleus)
Right
36
15
18
1
42
3
15 6
t-Value
51 24
6 6
39 24 12
36 42 48
12 48 0
33 15 18
81 51 51
3 18 18
0
36
51
Left
Right
3.23
3.55 3.38 3.27
2.96
3.07 3.01
3.24 3.23 3.06 2.89
BOLD signal changes were detected in the thalamus, insula, amygdala, primary and secondary somatosensory cortex, and in the middle and anterior cingulate cortex. These results are in line with
findings of a previous study of our group in healthy subjects using the same stimulation set-up [27]. However, an interaction analysis revealed that migraineurs significantly differ from control subjects in BOLD signal responses: In healthy subjects, the activity level increased during repetitive stimulation over time in several brain areas, whereas it decreased in migraine patients. These areas include the alPFC, rACC, the red nucleus and the ventral medulla most likely representing the olivary nucleus [18]. Most of these structures have been found to be involved in pain inhibitory circuits, attention modulation, cognitive evaluation, and planning motor behaviour in response to nociceptive input [22]. Specifically, the rACC is fundamentally involved in endogenous pain control, which is mediated by the opiodergic neurotransmitter system [21]. The PFC has been further interpreted to play a specific role in mediating attenuation of pain perception via cognitive control mechanisms [13,33]. Increased activation in the red nucleus as well as in the ventral medulla probably correspond to motor preparation and responses during pain sensation. Although the red nucleus is undoubtedly an important
102
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
relay station of a descending motor tract (rubrospinal tract), signal changes in this area have been linked to aversive conditioning [9] and peripheral noxious stimulation [5,14]. Moreover, electrical stimulation of the red nucleus modulates thalamic nociceptive transmission; [29] and frequent migraine has been reported to be associated with increased iron concentration/accumulation in this area [12]. Taken together, our longitudinal imaging data showed that migraineurs and controls significantly differ in a network of brain areas known to be fundamentally involved in pain inhibitory circuits [22]. The functional imaging data could be interpreted along the lines that migraine patients do have a dys-functional antinociceptive system. However, the behavioural data show no difference between migraineurs and controls, at least in our set-up (8 days of repetitive trigeminal pain). How can this obvious mismatch be explained? A general explanation would be that the read-out of the methods used (pain ratings and imaging data) are not sensitive enough to give us more detailed insight into the mechanisms of pain perception. Another possibility touches the methodological setup: based on previous work in healthy volunteers [7,28], we used an 8-day paradigm which resulted in robust behavioural and imaging data. It may well be that the involved neuronal systems exhaust over time and that this specific paradigm, 20 min trigeminal pain for eight consecutive days, is not sufficient to dismantle this behaviour. A third, more plausible explanation is that the relative extent of the activity level in neuronal structures involved in pain transmission and antinociceptive activity is strongly influenced by several factors. It may well be that inhibitory and/or facilitatory mechanisms, perhaps even at different levels of the trigeminal-thalamic-cortical pathway play a significant, yet unrecognized role. From a clinical view point, the hypothalamus would be the most likely ‘‘modulator” of the trigeminal pain system. Given the clinical progression of the migraine cycle, it is tempting to consider oscillating impulse generators in the limbic system, perhaps including the hypothalamus, which govern the activation level of the trigeminal nuclei but also antinociceptive networks towards an attack. Migraine is a cycling disorder characterized by recurrent headache attacks and various neuro-physiological symptoms. Several endogenous and environmental-related factors are known to trigger acute attacks. Consequently, diverse rhythms including endocrine, hormonal, and sleep rhythms may significantly influence the cortical excitability level in migraine patients over the migraine cycle [26]. This may just as well have a significant effect on the individual brain structures found to behave differently between migraine patients and controls. Furthermore, the distance to the next headache attack may be strongly correlated to the height of the neuronal activity level. Previous electrophysiological studies point to changes in the cortical excitability level in migraine patients during the pre-ictal state [1]. Due to the relatively long observation period (8 days), it was inevitable that the patients in our study were measured at different time points over their individual migraine cycle. It may well be that the attack distance or any other endogenous rhythm explains the significant variability in neuronal activations. A recent study found a cyclic thermal pain hypersensitivity to be present during the premonitory phase, but not earlier in the interictal period in migraineurs [24]. The authors suggested that hyperexcitable nociceptive sensory pathways might precede the migraine pain attack itself. We did not control for all possible scenarios, but excluded patients who suffered headache on a scanning day. This analysis did not change the results, suggesting that at least the headache attack per se had no influence on the overall findings. It is important to note that migraine patients significantly activated brain areas including the PFC and rACC during the first painful stimulation session on day one. Indeed, migraine patients activated
these structures in the first session significantly more than healthy controls. These data support the theory that the pain inhibitory network in migraineurs is not dys-functional per se. Quite the opposite, it seems rather that the brain response to a trigeminal nociceptive stimulus is relatively excessive compared to controls. However, the inverse behaviour of the regional brain areas (increasing activity in controls over time and decreasing activity in migraineurs) is not translated into behaviourally relevant levels, i.e. pain perception. This behaviour may mirror a general, hitherto unknown processing of repetitive nociceptive input which is inherent to migraine. Further studies, including chronic migraine, medication overuse headache and other primary headache disorders are necessary to understand this complex behaviour. In summary, this is the first study which investigated repetitive trigeminal-nociceptive stimulation in migraine and healthy controls using functional imaging. Although migraine patients and control subjects did not differ in pain perception over time, the activity level in certain antinociceptive structures such as the prefrontal cortex, the rostral part of the cingulate cortex, the red nucleus and the ventral medulla, increased in healthy subjects over time, whereas it decreased in migraine patients. Because migraine patients activated these structures in the first session significantly more than healthy controls, we do not believe that these data support a generally dys-functional pain inhibitory network in migraineurs. 5. Conflict of interest statement None of the authors has a conflict of interest regarding the contents of this manuscript. Acknowledgements This work was supported by grants from the DFG (MA 1862/23) and BMBF (371 57 01 & NeuroImageNord). This work was further supported by an unrestricted scientific grant from S.A. Almirall. References [1] Ambrosini A, Schoenen J. Electrophysiological response patterns of primary sensory cortices in migraine. J Headache Pain 2006;7:377–88. [2] Ambrosini A, de Noordhout AM, Sandor PS, Schoenen J. Electrophysiological studies in migraine: a comprehensive review of their interest and limitations. Cephalalgia 2003;23:13–31. [3] Aurora SK, Wilkinson F. The brain is hyperexcitable in migraine. Cephalalgia 2007;27:1442–53. [4] Aurora SK, Barrodale PM, Tipton RL, Khodavirdi A. Brainstem dysfunction in chronic migraine as evidenced by neurophysiological and positron emission tomography studies. Headache 2007;47:996–1003. [5] Bingel U, Quante M, Knab R, Bromm B, Weiller C, Buchel C. Subcortical structures involved in pain processing: evidence from single-trial fMRI. Pain 2002;99:313–21. [6] Bingel U, Schoell E, Herken W, Büchel C, May A. Habituation to painful stimulation involves the antinociceptive system. Pain 2007;131:21–30. [7] Bingel U, Herken W, Teutsch S, May A. Habituation to painful stimulation involves the antinociceptive system–a 1-year follow-up of 10 participants. Pain 2008;140:393–4. [8] Borsook D, Burstein R, Becerra L. Functional imaging of the human trigeminal system: opportunities for new insights into pain processing in health and disease. J Neurobiol 2004;61:107–25. [9] Büchel C, Morris J, Dolan RJ, Friston KJ. Brain systems mediating aversive conditioning: an event-related fMRI study. Neuron 1998;20:947–57. [10] Coppola G, Pierelli F, Schoenen J. Is the cerebral cortex hyperexcitable or hyperresponsive in migraine? Cephalalgia 2007;27:1427–39. [11] IHS: The International Classification of Headache Disorders. 2nd ed. Cephalalgia 2004;24(Suppl 1):9–160. [12] Kruit MC, Launer LJ, Overbosch J, van Buchem MA, Ferrari MD. Iron accumulation in deep brain nuclei in migraine: a population-based magnetic resonance imaging study. Cephalalgia 2009;29:351–9. [13] Lieberman MD, Jarcho JM, Berman S, Naliboff BD, Suyenobu BY, Mandelkern M, Mayer EA. The neuronal correlates of placebo effects: a disruption account. Neuroimage 2004;22:447–55.
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103 [14] Matsumoto RR, Walker JM. Inhibition of rubral neurons by noxious and nonnoxious pressure. Brain Res 1991;556:78–84. [15] May A. New insights into headache: an update on functional and structural imaging findings. Nat Rev Neurol 2009;5:199–209. [16] May A, Goadsby PJ. The trigemino-vascular system in humans: pathophysiologic implications for primary headache syndromes of the neural influences on the cerebral circulation. J Cereb Blood Flow Metab 1999;19:115–27. [17] Moulton EA, Burstein R, Tully S, Hargreaves R, Becerra L, Borsook D. Interictal dysfunction of a brainstem descending modulatory center in migraine patients. PLoS One 2008;3:e3799. [18] Naidich TP, Duvernoy HM, Delman BN, Sorensen AG, Kollias SS, Haacke EM. Duvernoy’s Atlas of the human brainstem and cerebellum. High-field MRI, surface anatomy, internal structure, vascularization and 3-D sectional anatomy. Wien: Springer-Verlag; 2009. [19] Nichols T, Brett M, Andersson J, Wager T, Poline JB. Valid conjunction inference with the minimum statistic. Neuroimage 2005;25:653–60. [20] Olesen J, Tfelt Hansen P, Welch K. The headaches. Philadelphia: Lippincott Williams & Wilkins; 2006. [21] Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia – imaging a shared neuronal network. Science 2002;295:1737–40. [22] Peyron R, Laurent B, Garcia-Larrea L. Functional imaging of brain responses to pain. A review and meta-analysis (2000). Neurophysiol Clin 2000;30:263–88. [23] Rennefeld C, Wiech K, Schoell ED, Lorenz J, Bingel U. Habituation to pain: further support for a central component. Pain 2010;148:503–8.
103
[24] Sand T, Zhitniy N, Nilsen KB, Helde G, Hagen K, Stovner LJ. Thermal pain thresholds are decreased in the migraine preattack phase. Eur J Neurol 2008;15:1199–205. [25] Sprenger T, Goadsby PJ. Migraine pathogenesis, state of pharmacological treatment options. BMC Med 2009;7:71. [26] Stankewitz A, May A. The phenomenon of changes in cortical excitability in migraine is not migraine-specific – a unifying thesis. Pain 2009;145:14–7. [27] Stankewitz A, Voit HL, Bingel U, Peschke C, May A. A new trigeminonociceptive stimulation model for event-related fMRI. Cephalalgia 2009. [28] Teutsch S, Herken W, Bingel U, Schoell E, May A. Changes in brain gray matter due to repetitive painful stimulation. Neuroimage 2008;42:845–9. [29] Wang Z, Wang J, Li X, Yuan Y, Fan G. Interleukin-1 beta of red nucleus involved in the development of allodynia in spared nerve injury rats. Exp Brain Res 2008;188:379–84. [30] Weiller C, May A, Limmroth V, Juptner M, Kaube H, Schayck RV, Coenen HH, Diener HC. Brain stem activation in spontaneous human migraine attacks. Nat Med 1995;1:658–60. [31] Welch KM, Nagesh V, Aurora SK, Gelman N. Periaqueductal gray matter dysfunction in migraine: cause or the burden of illness? Headache 2001;41:629–37. [32] Wessman M, Terwindt GM, Kaunisto MA, Palotie A, Ophoff RA. Migraine: a complex genetic disorder. Lancet Neurol 2007;6:521–32. [33] Wiech K, Ploner M, Tracey I. Neurocognitive aspects of pain perception. Trends Cognit Sci 2008;12:306–13.
PAIN 151 (2010) 97–103
www.elsevier.com/locate/pain
Neuronal mechanisms during repetitive trigemino-nociceptive stimulation in migraine patients David Aderjan 1, Anne Stankewitz 1, Arne May ⇑ Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Germany
a r t i c l e
i n f o
Article history: Received 12 February 2010 Received in revised form 10 May 2010 Accepted 19 June 2010
Keywords: Functional magnetic resonance imaging (fMRI) Migraine Habituation Trigemino-nociceptive stimulation Pain processing Endogenous pain control
a b s t r a c t Habituation deficits in various sensory modalities have been observed in migraine patients in several experimental designs. The underlying neuronal mechanisms are, however, still unknown. Past studies have used electrophysiological measures and focussed on habituation behaviour during one single session. We were interested in how repeated painful stimulation over several days is processed, perceived and modulated in migraineurs. Fifteen migraine patients and 15 healthy controls were stimulated daily with a 20 min trigeminal pain paradigm for eight consecutive days, using functional MRI performed on days one and eight and one follow-up measurement three months later. The results demonstrate that migraine patients did not differ in behavioural pain ratings compared to the controls at any time. However, functional imaging data revealed a significant difference in several brain areas over time. The activity level in the prefrontal cortex (PFC) and the rostral anterior cingulate cortex (rACC) increased in healthy control subjects from day one to day eight, whereas it decreased in migraine patients. These data suggest that several brain areas known to be involved in endogenous pain control show a completely opposite behaviour in migraine patients compared to healthy controls. These brain networks seem not to be disrupted per se in migraine patients but changed activity over time responding to repetitive nociceptive input. The alteration of pain inhibitory circuits may be the underlying mechanism responsible for the dys-functional neuronal filters of sensory input. Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Migraine is considered to be a genetically determined neurological disorder [32]. The characteristics, frequency and intensity of migraine attacks are, however, strongly affected by several endogenous and environmentally related factors [26]. Although acute headache is the predominant symptom in most cases, several other manifestations in the central nervous system such as photophobia, phonophobia, and nausea add to its complexity [20]. The pathogen mechanisms underlying migraine have been of intense research interest in the last century. There is strong evidence that several brain structures are fundamentally involved, including the trigemino-vascular system [8,16], the brainstem [4,17,30,31], and the cerebral cortex [10]. The role of the cortical excitability level in migraine disease has been studied extensively in a number of electrophysiological studies, using event-related potentials [10,25]. Most of these studies ⇑ Corresponding author. Address: Department of Systems Neuroscience, University of Hamburg-Eppendorf (UKE), Martinistrasse 52, D-20246 Hamburg, Germany. Tel.: +49 (0)40 74105 9189; fax: +49 (0)40 74105 9955. E-mail address: [email protected] (A. May). 1 These authors contributed equally to this work.
revealed that migraine patients suffer from a lack of habituation in various sensory modalities during their interictal migraine interval, whereas healthy control subjects habituated during the same period. Interestingly, in migraineurs amplitudes normalize just before and during acute headache attacks [2]. The underlying neuronal mechanisms are, however, still unclear and controversial [3,10]. A lack of habituation in migraineurs represents an altered cortical information-process, which has been hypothesized to be a consequence of the dys-functional neuronal inhibitory circuits [10]. Although habituation behaviour in migraine patients has been extensively explored in the last century, this has been done almost exclusively during one stimulation session, whereas little is known about the habituation behaviour in migraineurs over time. Most importantly, so far, no functional imaging data are available to shed light on the underlying neuronal mechanisms. This is different for healthy volunteers, where the physiological response to repetitive nociceptive input has been more fully investigated [6,7,23,28]. On the basis of these findings, we aimed to investigate migraine patients during daily trigemino-nociceptive stimulation on eight consecutive days and a follow-up measurement three months later. Behavioural pain ratings and BOLD (blood oxygen level dependence) signal changes during trigeminal pain processing
0304-3959/$36.00 Ó 2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2010.06.024
Ò
98
D. Aderjan et al. / PAIN 151 (2010) 97–103
were compared between migraineurs and healthy controls. Given the known genetically determined susceptibility of migraine patients, we hypothesized that these groups would differ in behavioural pain ratings over time and that the differences would be seen in the brain responses.
experiment. The subjects were instructed to breathe orally during the experiment to avoid fluctuations in stimulus concentrations due to respiratory airflow.
2. Materials and methods
The study phase consisted of an eight days program of daily trigemino-nociceptive stimulation and a follow-up measurement after 90 days. On day one, eight, and 90, the individuals were exposed to 15 trigemino-nociceptive stimuli (ammonia) and 15 non-noxious control stimuli (air puffs) during the event-related fMRI (Fig. 1a). The stimulation equipment and design used in the present study is described in detail elsewhere [27]. On days two to seven the stimulation was performed outside the scanner. The number of stimuli were increased to 25 for each condition (trigemino-nociceptive and controls stimuli). During these behavioural measurements which were performed in a separate room next to the scanner only behavioural intensity rating data were recorded. For intranasal stimulation an olfactometer was used which was validated in a previous study by our group [27]. This approach allows us to achieve a tolerable but effective trigeminal stimulation in the fMRI. Gaseous ammonia was applied intranasally through a thin tube and provoked a nociceptive perception in the receptive field of the first and second trigeminal nerve division. The experimental design and stimulation approach were described in detail elsewhere [27]. The software presentation (http://www.neurolabs.com, version 13.0) was used for controlling open states of specific magnet valves of the olfactometer and for programming the visual stimulation which was presented on a computer screen during the experiment. The stimulation procedure (Fig. 1b) consisted of the following events: Prior to each stimulus an attention task was implemented: After a jittered time period of 8–10 s a red fixation cross changed its colour to yellow. Participants were instructed to press a button on a button-box as soon as possible after this colour-change. Following another 8–10 s one of the two stimulus conditions (ammonia or air puff) was applied to the subjects for a period of 800 ms by opening the specific valves. The tube .was then neutralized by a following constant airflow. After a jittered time delay of 6–8 s individuals rated the intensity of perceived pain on a numerical rating scale (NRS) by operating a button-box with three fingers of the right hand. The NRS was presented as a vertical line of numbers from zero to ten (zero = no sensation; ten = unbearable pain). During the sessions outside the scanner (day 2–7) participants were instructed to verbalize ratings which were then logged in by the experimenter.
2.1. Subjects We scanned forty-five volunteers who were recruited locally. Participants gave written informed consent in accordance with the declaration of Helsinki and approved by the local Ethics Committee. Fifteen data sets had to be excluded due to pre-defined exclusion criteria regarding: clinical features (n = 3 patient has more than six attacks per month), not sufficient pain ratings (controls n = 1 and patients n = 2), technical problems with the scanner (controls n = 2 and patients n = 1), withdrawal from the experiment (patients n = 2), or movement artefacts (translation > 3 mm or rotation > 3 degrees in three axes) (controls n = 1 and patients n = 3). Complete data sets of fifteen migraineurs with and without aura and fifteen healthy control subjects were included in our statistical analyses. Demographical and clinical features are shown in Table 1. Migraine diagnosis was based on the classification of the International Headache Society [11]. None of the patients took any migraine prophylaxis. The patients were controlled for migraine attacks during the experiment, as well as for the attacks which occurred within 72 h both before and after the experiment. None of the participants had a history of internal, psychiatric or neurological disease, in particular no history of other headache or pain syndrome. We assessed any occurrence of minor pain events up to 4 weeks prior to the study period. The subjects and patients were instructed to take no medication during the study phase. The subjects were informed about the possibility to withdraw from the experiment at any time for any given reason. All the participants were remunerated for the participation. 2.2. Pre-experimental phase The participants were informed about the purpose of the study investigating the central nervous systems’ adaptive behaviour to repeated painful stimulation in the trigeminal receptive field in migraine patients. Prior to the fMRI experiment [27], the subjects were trained with this paradigm outside the scanner. Only those subjects who experienced a clearly painful sensation (which was pre-defined as a ‘‘4” or above on the NRS) following the administration of ammonia (pain condition) were included in the fMRI
2.3. Experimental protocol
2.4. Image acquisition Table 1 Demographical and clinical features and pain ratings on day one, eight and 90 are listed. Mean values of pain ratings are presented with the standard error of the mean (±sem). Migraineurs
Controls
Demography n (female) Mean age in years (±SD) Range of age
18 (9) 32.13 (9.95) 24–53
15 (8) 28.8 (7.73) 21–48
Clinical features Frequency of attacks Migraine with aura
1–6 6
Behavioural results Mean pain rating day 1 (±sem) Mean pain rating day 8 (±sem) Mean pain rating day 90 (±sem)
5.40 (0.32) 4.20 (0.34) 4.59 (0.38)
Imaging was conducted using a 3T Siemens Trio scanner with a 12-channel head coil. For functional scans, a gradient echo–echo planar imaging sequence (EPI) with TE = 30 ms and TR = 2620 ms was performed (voxel size = 3mm3, flip angle 80°, field of view 192 mm3). Each volume consisted of 42 axial slices with a thickness of 3 mm and a gap of 1 mm. High resolution T1-weighted structural images (voxel size = 1 mm3) were acquired additionally, using a MPRAGE sequence. 2.5. Analyses of behavioural data
5.96 (0.37) 4.37 (0.34) 4.37 (0.22)
Behavioural data were analysed using the Statistical Program for Social Sciences (SPSS; version 15.0; SPSS Inc., Chicago, IL, USA). The mean values and standard errors (m ± sem) of pain ratings were calculated for each measurement time point (days 1–8
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
99
Fig. 1. (a) Experimental design: trigemino-nociceptive stimulation was performed on eight consecutive days and on one follow-up three months later (day 90). Functional MRI was assessed on day one, eight and 90. (b) Shows the stimulation paradigm in detail which was performed on each experimental day. This paradigm consists of a reaction task, a stimulation exposure and a rating procedure on a numerical rating scale (NRS). After each trial, an inter-stimulus interval followed. On day one, eight and 90 the stimulation was performed during fMRI (15 stimuli were used for each of the two conditions). For behavioural measurements on day two to seven, which were recorded outside the scanner, the number of stimuli was increased to 25 as per condition.
and 90) separately for each group. Further, we compared the averaged pain ratings on days one, eight, and 90 (fMRI measurements), using a variance analysis (ANOVA) for repeated measures and tested for time and group effects as well as for an interaction between both. Three measurement points (day one, eight, and 90) were set as inner-subject factors (= time) whereas the betweensubject factor (group) split the sample into migraineurs (n = 15) and healthy subjects (n = 15). We determined whether the behavioural effect was primarily based on a reduction between the first eight days or on a lowering between days eight and 90, by setting the general-linear-model contrasts in ANOVA. We further calculated a linear regression analysis testing for significant effects over the study time course from day one to day eight for each group. For analyses of behavioural data (pain ratings), the statistical threshold was set to p < 0.05. 2.6. Image processing and statistical analysis Image processing and statistical analysis were conducted with Statistical Parametric Mapping (SPM5; Welcome Department for Imaging Neuroscience, London UK) running on matlab (Mathworks). The pre-processing included slice time correction, realignment, and spatial normalization on the Montreal Neurological Institute (MNI) stereotactic map. Finally, the data were smoothed using a 10 mm3 full-width at half-maximum (FWHM) isotopic Gaussian kernel. Analyses of single-subject data were performed using a general linear model approach (GLM). The first-level analysis of each subject consisted of the following regressors: (1) ammonia (= pain), (2) air puff, (3) reaction task (onset of the fixation cross until buttonbox press after colour changing), (4) rating procedure (onset of presenting NRS until rating was logged), and (5) movement parameters (translation and rotation around x-, y-, and z-axis). There was a jittered time interval of 10 s between the reaction task and the nociceptive input. This was done to demonstrate that the volunteers stayed alert and focussed on the task. Every single event was convolved with a canonical hemodynamic response function (HRF) with the time derivative, as implemented in SPM5. After estimation of the first-level design matrix, a linear contrast of
ammonia > baseline activity was defined. Separate contrast-images for each of the MR-recording on days one, eight, and 90 were generated, resulting in three images for each of the thirty subjects. Statistical parametric maps were interpreted by referring them to the probabilistic behaviour of Gaussian random fields. In order to analyze the group effects, contrast-images were inserted into a 2 3 flexible factorial design-matrix which consisted of two groups (controls and migraineurs) and three conditions (scan time points: day one, eight, and 90). (1) BOLD signal changes during trigemino-nociceptive stimulation during the first fMRI session in migraineurs and controls: To assess shared networks of pain-related activation in both the groups we calculated a conjunction analysis [19], including contrast-images for the ammonia > baseline activity of the first measurement point (day one) for each individual (n = 15 migraineurs and n = 15 controls). (2) Group comparison regarding BOLD signal changes over scan time points: We then performed an interaction analysis (flexible factorial design) between the group (migraineurs and controls) and the condition (time: day one, eight, and 90) testing whether migraineurs differ from controls in trigeminal pain processing over time. Contrasts weights were based on behavioural findings: decreased pain ratings were observed from day one to day eight in both groups while rating values on day 90 were comparable to those of day eight. In SPM, we tested for brain areas in which the activity increased in controls from day one to day eight and stayed constant at day 90 and decreased in migraineurs from day one to day eight and stayed constant at day 90 (see behavioural results and Fig. 3). Additionally, the opposite contrast was performed. We scanned patients at any time point during their interictal and ictal state. Due to the possible effects of acute headache on our outcome, we calculated an additional flexible 2 3-factorial design, excluding the contrast images of the patients scanned during acute head pain (n = 9). The statistical threshold for behavioural analyses were set to p < 0.05 (ANOVA for repeated measurements). The statistical threshold for the conjunction was set to p < 0.05 corrected for whole brain multiple comparisons, using familywise error (FWE) rate. For interaction analyses, we accepted an uncorrected threshold of p < 0.005 because we focussed our interest on small subcor-
Ò
100
D. Aderjan et al. / PAIN 151 (2010) 97–103
tical and brainstem structures which are highly challenging to detect [26]. 3. Results 3.1. Behavioural data Pain ratings: An ANOVA for repeated measures detected a statistically significant effect for the factor ‘‘time” (day one, eight, and 90) (F(1,13) = 15.47, p < 0.05) whereas no significant effect was revealed for the factor ‘‘group” (patients and controls) and no significant interaction effect between ‘‘group” and ‘‘time” was found. GLM (General-linear-model)-contrasts (ANOVA) revealed a significant reduction between day one and day eight, as well as between day one and day 90 (p < 0.05) (see Fig. 2). 3.2. Neuroimaging results BOLD response during trigemino-nociceptive stimulation on day one (the first stimulation session) in migraineurs and controls: A conjunction analysis revealed increased BOLD signals in several brain areas during trigemino-nociceptive stimulation, including bilateral activations in the amygdala (AMY), primary (SI) and secondary somatosensory cortex (SII), pallidum, the anterior and posterior portion of the insula cortex (IC), and in the operculum. Increased activity was further seen in the middle cingulate cortex (MCC), ipsilaterally in the cerebellum, and contralaterally in the supra-
(a)
* *
7 6 5 NRS
4 3 2
marginal and inferior frontal gyrus. MNI-coordinates of peak voxels are listed in Table 2. BOLD signal changes over time (day one, eight and 90): An interaction analysis (flexible factorial design) aiming to identify brain areas in which the activity during trigeminal pain stimulation increased in healthy controls from day one to day eight and stayed constant at day 90 and decreased in migraineurs from day one to day eight and stayed constant at day 90 revealed robust activation in the right anterior lateral prefrontal cortex (alPFC) (39, 36, 12; t(56) = 3.55), the rostral anterior cingulate cortex (rACC) (12, 48, 0; t(56) = 3.27), the midbrain at the anatomical localization of the red nucleus ( 6, 24, 6; t(56) = 3.01), and the ventral part of the upper medulla oblongata (0, 36, 51; t(56) = 2.89). Additionally, increased BOLD responses were detected in the right occipital (33, 81, 3; t(56) = 3.24) and frontal cortex (15, 51, 18; t(56) = 3.23), in the right cerebellum (24, 42, 48; t(56) = 3.38), and bilaterally in the calcarine gyrus ( 15, 51, 6; t(56) = 3.07 and 18, 48, 9; t(56) = 3.06) (p < 0.005 uncorrected) (peak voxel of activation patterns are listed in X-, Y- and Z-MNI space). The opposite contrast (areas in which activity increased in migraineurs and decreased in controls during trigeminal pain stimulation from day one to day eight and stayed constant at day 90) did not reveal any significant result (p < 0.005 uncorrected). Some patients (n = 9) reported acute headache on one of the scanning days. This is to be expected, given that the study covered 3 scanning days in a total of 90 days. We therefore used additionally a flexible factorial design which excluded these patients. This analysis did not change the results and revealed the same activation pattern as the interaction analysis including all contrast images. To exclude the possibility that patients and controls differ regarding attention, we performed an additional analysis, comparing linear contrasts modelling the activity during the reaction task at day 1 between both groups. No significant differences could be seen between migraineurs and controls (whole brain analysis; uncorrected; p < 0.005). Furthermore, reaction time scores did not explain any differences between groups and the height of the reaction time score did not correlate with the height of the BOLD signal intensity during the nociceptive stimulation neither in controls nor in migraine patients (whole brain analysis; uncorrected; p < 0.005).
1 0
4. Discussion
90
8
1
Experimental days
(b)
7 6 5
NRS
4 3 controls
2
migraineurs 1 0 1
2
3
4
5
6
7
8
Experimental days Fig. 2. The upper part of the figure (a) shows mean values of pain ratings (NRS) of healthy control subjects (black bar) and of migraine patients (white bar) on the days of scanning. A significant attenuation of pain ratings from day one to eight was seen in both groups. The pain ratings on day 90 were comparable with those on day eight. The lower part of the figure (b) shows the averaged pain ratings (NRS values) on each single day for healthy subjects (grey bar) and migraineurs (blue bar). Pain ratings showed a gradual decrease over time in both groups. Error bars indicate the standard error of the mean (sem) (see Table 1).
Contrary to our expectation that migraine patients and control subjects would differ in pain perception over time, both groups behaved the same during the eight experimental days. Migraine patients and healthy controls showed a gradual decrease of pain ratings from day one to day eight which only marginally increased again on day 90. This behaviour is identical to several studies in healthy controls using repetitive nociceptive input [6,7,23,28]. Whatever the neuronal basis for intra-session habituation may be [3,10], repeated nociceptive input over several sessions does not result in a loss of habituation in migraineurs. Of note, the neurophysiological evidences of a reduced habituation in migraine patients were issued from experiments of repetitive sensory stimuli during a single session with a very short inter-stimulus interval. By contrast, our study focussed on habituation behaviour over several days. It is therefore more than likely that the neurophysiological mechanisms behind both effects are fundamentally different. We further focussed on functional MRI data demonstrating that both groups activated a pattern of brain areas known to be involved in sensory, affective, and cognitive pain processing during the first trigeminal pain stimulation on day one [15]. Increased
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
101
Fig. 3. The interaction analysis between migraine patients and controls was contrasted to identify brain areas in which the brain activity in healthy controls during trigeminal nociceptive stimulation showed a significant increase from day one to eight (which stayed constant on day 90) whereas the activity level in these areas decreased in migraine patients. Significant BOLD signal changes were detected in the prefrontal cortex (PFC), in the rostral anterior cingulate cortex (rACC), in the midbrain (red nucleus) (NR) and in the ventral medulla. Results are plotted onto the averaged template of high resolution structural images of healthy controls and interictal migraine patients.
Table 2 MNI coordinates of the conjunction analysis of the first scanning day of migraineurs and controls. Results are FWE-corrected (p < 0.05). Anatomical structure
Coordinates of peak voxel (x, y, z in mm) Left
Prefrontal cortex (PFC) Cerebellum Rostral anterior cingulate cortex (rACC) Occipital cortex Frontal cortex Calcarine gyrus Nucleus ruber Ventral medulla (olivary nucleus)
Right
36
15
18
1
42
3
15 6
t-Value
51 24
6 6
39 24 12
36 42 48
12 48 0
33 15 18
81 51 51
3 18 18
0
36
51
Left
Right
3.23
3.55 3.38 3.27
2.96
3.07 3.01
3.24 3.23 3.06 2.89
BOLD signal changes were detected in the thalamus, insula, amygdala, primary and secondary somatosensory cortex, and in the middle and anterior cingulate cortex. These results are in line with
findings of a previous study of our group in healthy subjects using the same stimulation set-up [27]. However, an interaction analysis revealed that migraineurs significantly differ from control subjects in BOLD signal responses: In healthy subjects, the activity level increased during repetitive stimulation over time in several brain areas, whereas it decreased in migraine patients. These areas include the alPFC, rACC, the red nucleus and the ventral medulla most likely representing the olivary nucleus [18]. Most of these structures have been found to be involved in pain inhibitory circuits, attention modulation, cognitive evaluation, and planning motor behaviour in response to nociceptive input [22]. Specifically, the rACC is fundamentally involved in endogenous pain control, which is mediated by the opiodergic neurotransmitter system [21]. The PFC has been further interpreted to play a specific role in mediating attenuation of pain perception via cognitive control mechanisms [13,33]. Increased activation in the red nucleus as well as in the ventral medulla probably correspond to motor preparation and responses during pain sensation. Although the red nucleus is undoubtedly an important
102
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103
relay station of a descending motor tract (rubrospinal tract), signal changes in this area have been linked to aversive conditioning [9] and peripheral noxious stimulation [5,14]. Moreover, electrical stimulation of the red nucleus modulates thalamic nociceptive transmission; [29] and frequent migraine has been reported to be associated with increased iron concentration/accumulation in this area [12]. Taken together, our longitudinal imaging data showed that migraineurs and controls significantly differ in a network of brain areas known to be fundamentally involved in pain inhibitory circuits [22]. The functional imaging data could be interpreted along the lines that migraine patients do have a dys-functional antinociceptive system. However, the behavioural data show no difference between migraineurs and controls, at least in our set-up (8 days of repetitive trigeminal pain). How can this obvious mismatch be explained? A general explanation would be that the read-out of the methods used (pain ratings and imaging data) are not sensitive enough to give us more detailed insight into the mechanisms of pain perception. Another possibility touches the methodological setup: based on previous work in healthy volunteers [7,28], we used an 8-day paradigm which resulted in robust behavioural and imaging data. It may well be that the involved neuronal systems exhaust over time and that this specific paradigm, 20 min trigeminal pain for eight consecutive days, is not sufficient to dismantle this behaviour. A third, more plausible explanation is that the relative extent of the activity level in neuronal structures involved in pain transmission and antinociceptive activity is strongly influenced by several factors. It may well be that inhibitory and/or facilitatory mechanisms, perhaps even at different levels of the trigeminal-thalamic-cortical pathway play a significant, yet unrecognized role. From a clinical view point, the hypothalamus would be the most likely ‘‘modulator” of the trigeminal pain system. Given the clinical progression of the migraine cycle, it is tempting to consider oscillating impulse generators in the limbic system, perhaps including the hypothalamus, which govern the activation level of the trigeminal nuclei but also antinociceptive networks towards an attack. Migraine is a cycling disorder characterized by recurrent headache attacks and various neuro-physiological symptoms. Several endogenous and environmental-related factors are known to trigger acute attacks. Consequently, diverse rhythms including endocrine, hormonal, and sleep rhythms may significantly influence the cortical excitability level in migraine patients over the migraine cycle [26]. This may just as well have a significant effect on the individual brain structures found to behave differently between migraine patients and controls. Furthermore, the distance to the next headache attack may be strongly correlated to the height of the neuronal activity level. Previous electrophysiological studies point to changes in the cortical excitability level in migraine patients during the pre-ictal state [1]. Due to the relatively long observation period (8 days), it was inevitable that the patients in our study were measured at different time points over their individual migraine cycle. It may well be that the attack distance or any other endogenous rhythm explains the significant variability in neuronal activations. A recent study found a cyclic thermal pain hypersensitivity to be present during the premonitory phase, but not earlier in the interictal period in migraineurs [24]. The authors suggested that hyperexcitable nociceptive sensory pathways might precede the migraine pain attack itself. We did not control for all possible scenarios, but excluded patients who suffered headache on a scanning day. This analysis did not change the results, suggesting that at least the headache attack per se had no influence on the overall findings. It is important to note that migraine patients significantly activated brain areas including the PFC and rACC during the first painful stimulation session on day one. Indeed, migraine patients activated
these structures in the first session significantly more than healthy controls. These data support the theory that the pain inhibitory network in migraineurs is not dys-functional per se. Quite the opposite, it seems rather that the brain response to a trigeminal nociceptive stimulus is relatively excessive compared to controls. However, the inverse behaviour of the regional brain areas (increasing activity in controls over time and decreasing activity in migraineurs) is not translated into behaviourally relevant levels, i.e. pain perception. This behaviour may mirror a general, hitherto unknown processing of repetitive nociceptive input which is inherent to migraine. Further studies, including chronic migraine, medication overuse headache and other primary headache disorders are necessary to understand this complex behaviour. In summary, this is the first study which investigated repetitive trigeminal-nociceptive stimulation in migraine and healthy controls using functional imaging. Although migraine patients and control subjects did not differ in pain perception over time, the activity level in certain antinociceptive structures such as the prefrontal cortex, the rostral part of the cingulate cortex, the red nucleus and the ventral medulla, increased in healthy subjects over time, whereas it decreased in migraine patients. Because migraine patients activated these structures in the first session significantly more than healthy controls, we do not believe that these data support a generally dys-functional pain inhibitory network in migraineurs. 5. Conflict of interest statement None of the authors has a conflict of interest regarding the contents of this manuscript. Acknowledgements This work was supported by grants from the DFG (MA 1862/23) and BMBF (371 57 01 & NeuroImageNord). This work was further supported by an unrestricted scientific grant from S.A. Almirall. References [1] Ambrosini A, Schoenen J. Electrophysiological response patterns of primary sensory cortices in migraine. J Headache Pain 2006;7:377–88. [2] Ambrosini A, de Noordhout AM, Sandor PS, Schoenen J. Electrophysiological studies in migraine: a comprehensive review of their interest and limitations. Cephalalgia 2003;23:13–31. [3] Aurora SK, Wilkinson F. The brain is hyperexcitable in migraine. Cephalalgia 2007;27:1442–53. [4] Aurora SK, Barrodale PM, Tipton RL, Khodavirdi A. Brainstem dysfunction in chronic migraine as evidenced by neurophysiological and positron emission tomography studies. Headache 2007;47:996–1003. [5] Bingel U, Quante M, Knab R, Bromm B, Weiller C, Buchel C. Subcortical structures involved in pain processing: evidence from single-trial fMRI. Pain 2002;99:313–21. [6] Bingel U, Schoell E, Herken W, Büchel C, May A. Habituation to painful stimulation involves the antinociceptive system. Pain 2007;131:21–30. [7] Bingel U, Herken W, Teutsch S, May A. Habituation to painful stimulation involves the antinociceptive system–a 1-year follow-up of 10 participants. Pain 2008;140:393–4. [8] Borsook D, Burstein R, Becerra L. Functional imaging of the human trigeminal system: opportunities for new insights into pain processing in health and disease. J Neurobiol 2004;61:107–25. [9] Büchel C, Morris J, Dolan RJ, Friston KJ. Brain systems mediating aversive conditioning: an event-related fMRI study. Neuron 1998;20:947–57. [10] Coppola G, Pierelli F, Schoenen J. Is the cerebral cortex hyperexcitable or hyperresponsive in migraine? Cephalalgia 2007;27:1427–39. [11] IHS: The International Classification of Headache Disorders. 2nd ed. Cephalalgia 2004;24(Suppl 1):9–160. [12] Kruit MC, Launer LJ, Overbosch J, van Buchem MA, Ferrari MD. Iron accumulation in deep brain nuclei in migraine: a population-based magnetic resonance imaging study. Cephalalgia 2009;29:351–9. [13] Lieberman MD, Jarcho JM, Berman S, Naliboff BD, Suyenobu BY, Mandelkern M, Mayer EA. The neuronal correlates of placebo effects: a disruption account. Neuroimage 2004;22:447–55.
Ò
D. Aderjan et al. / PAIN 151 (2010) 97–103 [14] Matsumoto RR, Walker JM. Inhibition of rubral neurons by noxious and nonnoxious pressure. Brain Res 1991;556:78–84. [15] May A. New insights into headache: an update on functional and structural imaging findings. Nat Rev Neurol 2009;5:199–209. [16] May A, Goadsby PJ. The trigemino-vascular system in humans: pathophysiologic implications for primary headache syndromes of the neural influences on the cerebral circulation. J Cereb Blood Flow Metab 1999;19:115–27. [17] Moulton EA, Burstein R, Tully S, Hargreaves R, Becerra L, Borsook D. Interictal dysfunction of a brainstem descending modulatory center in migraine patients. PLoS One 2008;3:e3799. [18] Naidich TP, Duvernoy HM, Delman BN, Sorensen AG, Kollias SS, Haacke EM. Duvernoy’s Atlas of the human brainstem and cerebellum. High-field MRI, surface anatomy, internal structure, vascularization and 3-D sectional anatomy. Wien: Springer-Verlag; 2009. [19] Nichols T, Brett M, Andersson J, Wager T, Poline JB. Valid conjunction inference with the minimum statistic. Neuroimage 2005;25:653–60. [20] Olesen J, Tfelt Hansen P, Welch K. The headaches. Philadelphia: Lippincott Williams & Wilkins; 2006. [21] Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia – imaging a shared neuronal network. Science 2002;295:1737–40. [22] Peyron R, Laurent B, Garcia-Larrea L. Functional imaging of brain responses to pain. A review and meta-analysis (2000). Neurophysiol Clin 2000;30:263–88. [23] Rennefeld C, Wiech K, Schoell ED, Lorenz J, Bingel U. Habituation to pain: further support for a central component. Pain 2010;148:503–8.
103
[24] Sand T, Zhitniy N, Nilsen KB, Helde G, Hagen K, Stovner LJ. Thermal pain thresholds are decreased in the migraine preattack phase. Eur J Neurol 2008;15:1199–205. [25] Sprenger T, Goadsby PJ. Migraine pathogenesis, state of pharmacological treatment options. BMC Med 2009;7:71. [26] Stankewitz A, May A. The phenomenon of changes in cortical excitability in migraine is not migraine-specific – a unifying thesis. Pain 2009;145:14–7. [27] Stankewitz A, Voit HL, Bingel U, Peschke C, May A. A new trigeminonociceptive stimulation model for event-related fMRI. Cephalalgia 2009. [28] Teutsch S, Herken W, Bingel U, Schoell E, May A. Changes in brain gray matter due to repetitive painful stimulation. Neuroimage 2008;42:845–9. [29] Wang Z, Wang J, Li X, Yuan Y, Fan G. Interleukin-1 beta of red nucleus involved in the development of allodynia in spared nerve injury rats. Exp Brain Res 2008;188:379–84. [30] Weiller C, May A, Limmroth V, Juptner M, Kaube H, Schayck RV, Coenen HH, Diener HC. Brain stem activation in spontaneous human migraine attacks. Nat Med 1995;1:658–60. [31] Welch KM, Nagesh V, Aurora SK, Gelman N. Periaqueductal gray matter dysfunction in migraine: cause or the burden of illness? Headache 2001;41:629–37. [32] Wessman M, Terwindt GM, Kaunisto MA, Palotie A, Ophoff RA. Migraine: a complex genetic disorder. Lancet Neurol 2007;6:521–32. [33] Wiech K, Ploner M, Tracey I. Neurocognitive aspects of pain perception. Trends Cognit Sci 2008;12:306–13.