Physiological brainstem mechanisms of trigeminal nociception: An fMRI study at 3T

Physiological brainstem mechanisms of trigeminal nociception: An fMRI study at 3T

NeuroImage 124 (2016) 518–525 Contents lists available at ScienceDirect NeuroImage journal homepage: www.elsevier.com/locate/ynimg Physiological br...

1MB Sizes 0 Downloads 55 Views

NeuroImage 124 (2016) 518–525

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

Physiological brainstem mechanisms of trigeminal nociception: An fMRI study at 3T Laura H. Schulte, Christian Sprenger, Arne May ⁎ Department of Systems Neuroscience, Center for Experimental Medicine, University Medical Center Eppendorf, Martinistr. 52, 20246 Hamburg, Germany

a r t i c l e

i n f o

Article history: Received 20 April 2015 Accepted 10 September 2015 Available online 21 September 2015 Keywords: Brainstem fMRI Trigeminus Pain Nociception Headache

a b s t r a c t The brainstem is a major site of processing and modulation of nociceptive input and plays a key role in the pathophysiology of various headache disorders. However, human imaging studies on brainstem function following trigeminal nociceptive stimulation are scarce as brainstem specific imaging approaches have to address multiple challenges such as magnetic field inhomogeneities and an enhanced level of physiological noise. In this study we used a viable protocol for brainstem fMRI of standardized trigeminal nociceptive stimulation to achieve detailed insight into physiological brainstem mechanisms of trigeminal nociception. We conducted a study of 21 healthy participants using a nociceptive ammonia stimulation of the left nasal mucosa with an optimized MR acquisition protocol for high resolution brainstem echoplanar imaging in combination with two different noise correction techniques. Significant BOLD responses to noxious ammonia stimulation were observed in areas typically involved in trigeminal nociceptive processing such as the spinal trigeminal nuclei (sTN), thalamus, secondary somatosensory cortex, insular cortex and cerebellum as well as in a pain modulating network including the periaqueductal gray area, hypothalamus (HT), locus coeruleus and cuneiform nucleus (CNF). Activations of the left CNF were positively correlated with pain intensity ratings. Employing psychophysiological interaction (PPI) analysis we found enhanced functional connectivity of the sTN with the contralateral sTN and HT following trigeminal nociception. We also observed enhanced functional connectivity of the CNF with the RVM during painful stimulation thus implying an important role of these two brainstem regions in central pain processing. The chosen approach to study trigeminal nociception with high-resolution fMRI offers new insight into human pain processing and might thus lead to a better understanding of headache pathophysiology. © 2015 Published by Elsevier Inc.

Introduction The brainstem is the major site of trigeminal pain processing and modulation and has been discussed as a main player in the pathophysiology of various headache disorders such as migraine or cluster headache (Afridi et al., 2005; Bahra et al., 2001; Denuelle et al., 2007; Stankewitz et al., 2011; Weiller et al., 1995). Previous anatomical and electrophysiological studies in animals and humans have been able to identify the primary hubs of the trigeminal pain pathway: a nociceptive stimulus enters the central nervous system via the trigeminal ganglion (TG). The stimulus is then conducted to the spinal trigeminal nuclei (sTN) and further to the ventral posterior medial nucleus of the thalamus (VPM) and to the primary somatosensory cortex (S1) (Wallenberg, 1900; Carpenter and Hanna, 1961; Stewart and King, 1963; Tiwari and King, 1974; Hayashi, 1985a, 1985b; Capra and Dessem, 1992; Usunoff et al., 1997; DaSilva et al., 2002). ⁎ Corresponding author at: Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany. Fax: +49 40 7410 59955. E-mail address: [email protected] (A. May).

http://dx.doi.org/10.1016/j.neuroimage.2015.09.023 1053-8119/© 2015 Published by Elsevier Inc.

Nociceptive signals are modulated by a network of various brainstem centers: the descending pain modulating system, including the periaqueductal gray matter (PAG), the hypothalamus (HT), the red nucleus (RN), the substantia nigra (SN), the locus coeruleus (LC), the rostral ventromedial medulla and the lateral reticular formation of the midbrain, including the cuneiform nucleus (CNF) (Millan, 2002). Whereas the anatomy and fiber connections between the aforementioned brainstem centers are reasonably well understood from anatomical and electrophysiological work (Basbaum et al., 2009; Burstein et al., 1998; Edvinsson, 2011; Millan, 2002; Noseda et al., 2008; Noseda and Burstein, 2013), their functional activity and their interactions in humans following trigeminal nociceptive stimulation remain a subject of current research. Previous functional imaging studies in humans have been able to identify the spinal trigeminal nuclei, the VPM and SI as the major conduction sites of painful trigeminal stimuli (DaSilva et al., 2002; Stankewitz et al., 2010). However, the aforementioned studies commonly investigated both cortical and subcortical areas without a special focus on the brainstem. As a consequence, brainstem activations following trigeminal nociceptive stimulation in most studies could either not be found at all (Boyle et al., 2007; Iannilli et al., 2008) or have been very unspecific without

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

519

a clear differentiation between different brainstem nuclei (Hummel et al., 2005). Thus a more focused imaging technique of the human brainstem in combination with human models of headache attacks is an indispensable tool in pathophysiologic and therapeutic headache research. However, as the brainstem is an area directly surrounded by large blood vessels and cerebrospinal fluid, it is much more susceptible to cardiac and respiratory noise than other regions of the brain. Additionally, as the different nuclei of this brain area lie in a close spatial relationship to each other, a high spatial resolution is mandatory to assign found activations to single brainstem nuclei. This however decreases the signal to noise ratio (SNR). As a result a special noise correction technique is mandatory. To address these points and to achieve a more detailed insight into mechanisms of trigeminal nociception within the brainstem, we employed a well-established protocol of nociceptive trigeminal stimulation in combination with an optimized approach to brainstem fMRI using an adapted field of view and a high in-plane resolution in combination with retrospective correction of general image noise by spatially adaptive nonlocal means filtering (Coupé et al., 2006) and retrospective correction of physiological noise (Deckers et al., 2006). We present detailed insight into brainstem mechanisms of trigeminal nociception.

presented using a custom built olfactometer. Via this device, low concentrated gaseous ammonia (2.5% ammonia concentration in the air reaching the left nostril) or rose odor embedded in a constant airflow of 74.4 ml/s was administered to the left nostril of each participant for a duration of 0.8 s via an 8 m Teflon tube. Visual stimuli as well as the reaction task and the rating bars for the stimuli were presented via a mirror system using Presentation® software (version 16.4, Neurobehavioral Systems, Berkeley, California, USA). Every stimulus was presented 15 times during the whole experiment, divided into three sessions. Stimuli followed each other in a pseudorandomized order thus ensuring that no two adjacent stimuli were of the same kind. During each trial participants were not told which stimulus was to come. Prior to each stimulus a reaction task was performed: subjects were instructed to press a button on a button box when a white cross turned red. Following each stimulus presentation, participants were asked to rate the intensity of stimuli on a visual analogue scale from 0 to 100, as well as the pleasantness of each stimulus (− 50: very pleasant, 0: neutral, +50: very unpleasant). See Fig. 1 for further detail. Before the experiment conduction in the scanner, participants underwent a short training session. Special care was taken to explain the meaning of the two scales and the corresponding ratings.

Methods

Behavioral data

Subjects

Intensity and pleasantness of all stimuli were rated on a visual numeric rating scale. Mean and 95%-confidential intervals (95%-CIs) of intensity and pleasantness ratings were calculated using Matlab (R2013b, The MathWorks, Inc., Natick, Massachusetts, USA).

24 healthy subjects (age 29 ± 1.2 years, 13 female, 4 male) were scanned for this study. Exclusion criteria were a previous diagnosis of any headache or facial pain disorder, severe psychiatric, neurological or internal illness, pregnancy, lactation and general safety concerns regarding the participants' ability to undergo MRI examination and/or painful ammonia stimulation. Three subjects had to be excluded post hoc for major susceptibility artifacts or corrupted data. 21 participants were included in the analyses. Study design To study the mechanisms of trigeminal nociceptive processing we used a well established protocol for standardized trigeminal nociceptive stimulation that has in detail been described in previous publications (Stankewitz et al., 2010). To avoid tactile sensation on the skin and because all the material used in the high-magnetic-field-room has to be non-ferrous, we used an olfactometer. The paradigm basically consisted of 4 different stimuli: Gaseous ammonia as an activator of trigeminal nociceptive fibers and medullary dorsal horn neurons (McCulloch et al., 2013; Stankewitz et al., 2010) served as the painful stimulus, rose odor as the olfactory stimulus, air as control condition and a rotating checkerboard as the visual stimulus. The three gaseous stimuli were

MRI data acquisition MRI images were acquired on a Siemens TIM TRIO 3T system equipped with a 32-channel head coil. All participants lay supine on the back in the scanner. To minimize head movements, foam pads were inserted on either side of the participants' head. Special attention was paid to a straight and standardized positioning of the head inside the head coil. We used an image acquisition protocol that was specifically adapted for high resolution brainstem echoplanar imaging (voxel sizes 1.25 × 1.25 × 2.5 mm3, TR 2.61 s, TE 27 ms, 38 axial slices, FoV 216 × 108 mm2, matrix: 172 × 86, flip angle 80°, parallel accelerated with GRAPPA (PAT factor 2, 48 reference lines), 2 spatial saturation pulses were applied anterior and posterior to the target volume). To obtain a reasonable SNR despite realizing a high in-plane resolution, we increased the slice thickness to 2.5 mm. With respect to the columnar structure of many brainstem nuclei we judged this to be reasonable. Given the reduced field of view the application of saturation pulses anterior and posterior to the target volume was reasonable to minimize ghosting and aliasing artifacts. Susceptibility artifacts were minimized

Fig. 1. Schematic overview of the experimental paradigm. Each trial is led by a reaction task (button press when white cross turns red) which is then followed by one of the four different stimuli (ammonia, rose odor, air, checkerboard). After a jittered time interval of 8 to 10 s, subjects were asked to rate intensity and pleasantness of the preceding stimulus. Each stimulus was presented 15 times over the whole experiment in a pseudorandomized order.

520

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

by performing a 3d-shim with three iterations using the shim algorithm provided by the manufacturer and a manually defined shim volume confined to the brainstem at the beginning of each experiment. The foramen magnum in a median sagittal slice served as a reference point for alignment of the lowest slice of each volume. Finally, high resolution T1 weighted images were recorded using an MPRAGE sequence (voxel size 1 × 1 × 1 mm3, TR 2.3 s, TE 2.98 ms, FoV 192 × 256 × 240 mm3, slice orientation: sagittal, flip angle 9°, inversion time 1.1 s). Nonlocal means filtering and image preprocessing Image processing and statistical analysis of imaging data were performed using SPM8 (Wellcome Department for Imaging Neuroscience, London, UK). The first five volumes of each time series were discarded to exclude T1 saturation effects. Given the lower SNR than in whole brain MRI approaches due to the higher in-plane resolution, the anatomical position of the brainstem and its small size (Harvey et al., 2008), we applied a special correction technique to reduce total image noise. Nonlocal means algorithms provide a viable approach to disentangle image noise from relevant image information. This approach restores the intensity value of a given voxel in a defined search volume using an average of intensities of all voxels in this search volume weighted according to the distortion between voxel neighborhood intensities and has been shown to outperform other established noise correction techniques (Coupé et al., 2006). Prior to the classical preprocessing steps echoplanar images were thus filtered using a nonlocal means algorithm as implemented in the spatially adaptive nonlocal means denoising filter of the SPM toolbox VBM8. After slice time correction, functional images were realigned using a standard rigid body transformation with six degrees of freedom as implemented in SPM8. Images were then normalized into MNI space via the following steps: first, the high resolution structural T1 weighted image of each person was coregistered to the mean functional image of the same subject and then normalized using an iterative segmentation–normalization approach (Ashburner and Friston, 2005). The obtained normalization parameters were then applied to the functional images and functional images were upsampled to a resolution of 1 × 1 × 1 mm3. Correctness of normalization was checked for each subject by inspection. Functional images were then smoothed using a 4 mm3 full width at half maximum (FWHM) isotropic Gaussian kernel to facilitate group analysis, but at the same time to allow an appropriate anatomical assignment within the brainstem. This smoothing kernel dimension has previously been validated as feasible and recommendable for group analysis of brainstem fMRI data (Beissner et al., 2011). Physiological noise correction As the brainstem due to its anatomical location is an area susceptible to cardiac and respiratory noise, a special correction technique is mandatory (Beissner, 2015; Brooks et al., 2013; Harvey et al., 2008). We thus recorded cardiac (pulse curve recorded via a pulsoxymeter on the left index finger) and respiratory signals together with the scanner pulses using an MRI compatible monitoring system (Expression, InVivo, Gainesville, USA). Cardiac and respiratory data as well as the scanner pulses were recorded via a CED 1401 Micro3 device and Spike2 software (Cambridge Electronic Design, Cambridge, UK). To correct for physiological noise we followed the selective averaging approach described by Deckers and colleagues (Deckers et al., 2006). In short, each individual EPI volume was attributed to 10 bins of the respiratory and the cardiac cycle, respectively, thus creating cardiac and respiratory noise regressors to be included in the 1st level general linear model. Despite the recent critique of this method by Kong et al. (Kong et al., 2012) regarding the application of this method in functional imaging of the spinal cord, it has been shown to be at least equal to the long established RETROICOR (Deckers et al., 2006).

Statistical fMRI analysis Statistical analysis was performed using the general linear model approach as implemented in SPM8. Six experimental regressors were included in the first level design matrix: ammonia (1), rose odor (2), air (3), checkerboard (4), anticipation phase (5), and button presses (6). Respiratory and cardiac noise regressors as well as six movement parameters (three translational and three rotational) were included in the analysis. As an anticipation interval we defined the pause between the reaction task and the onset of each of the four stimuli. The experimental regressors ammonia, rose odor, air and button presses were modeled by convolving a delta function at event onset with the canonical hemodynamic response function (HRF). For the other two stimuli (checkerboard, anticipation phase) box car functions of short durations (4 s for checkerboard and 4–6 s for anticipation phase) were convolved with the HRF.

Parametric modulation To test for the linear effect of pain ratings, we set up a second first level analysis additionally including pain intensity ratings as parametric modulator of the stimulus response function with a 1st degree polynomial expansion. In both cases, first level contrast images were defined for the main effect of pain intensity.

Functional connectivity As we were particularly interested in functional connectivity of the left sTN as the first site of trigeminal pain processing within the central nervous system, we conducted a psychophysiological interaction analysis (Friston et al., 1997) by using the left sTN as the seed region. As the sTN is an elongated structure that cannot satisfactorily be represented by a sphere, we created a mask in the shape of the main activity in the sTN following trigeminonociceptive stimulation as seen in the second level analysis at a threshold of p b 0.00001 for the pain contrast (ammonia N air) using MarsBar toolbox. This threshold was chosen for ROI-definition as it provided a clear cluster in accordance with the anatomical extent of the left sTN. This mask was then used for seed region definition. As we identified the CNF as specifically related to perceived pain intensity, we additionally conducted a PPI analysis using this region as the seed, based on a 3 mm sphere around the peak voxel of the active cluster from the parametric modulation analysis (see above). Thus three regressors (1. the extracted time course from the seed region, 2. painful ammonia stimulation versus air as the psychological/stimulus dimension and 3. the PPI term, i.e. time series × stimulus variable) plus cardiac, respiratory and movement regressors were included in a new first level GLM analysis.

Statistical tests performed at the group level To test for shared activations of all subjects following trigeminal nociceptive, olfactory or visual stimulation, we performed onesample t tests using the pain-contrast-images (ammonia N air), odor-contrast-images (rose odor N air) and the main effect of checkerboard stimulation. The contrast to air stimulation was chosen for the nasally applied stimuli to control for effects of air flow and other tactile sensations due to the mere presence of the tube. Regarding parametric modulation and PPI analyses, also one-sample t tests were performed for the main effect of pain intensity in case of the parametric modulation and for the interaction term in case of PPI analyses.

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

Correction for multiple comparisons Results were considered significant at p b 0.05, corrected for multiple comparisons using the family-wise error (FWE) rate. Based on the preexisting neurophysiological knowledge and results from previous studies, we defined several a priori regions of interest. In the brainstem, the medulla, pons and mesencephalon and in the brain the left and right thalamus, insula, SII and amygdala were considered as a priori regions of interest. Given their well-known role during pain processing we also defined more specifically the sTN, RVM, HT, LC and the CNF as a priori regions of interest. For these regions the multiple comparison correction was restricted to 3-dimensional structural masks. If possible masks were obtained from the Wake Forest University PickAtlas (WFUPickAtlas, release 3.0.4, (Maldjian et al., 2003)). As for smaller brainstem regions no structural masks are currently available we used previously reported coordinates with 6 mm spheres for these regions (HT (May et al., 1998), CNF (Zwergal et al., 2013), RVM (coordinates (x, y, z): 0 -34 -50) (Vachon-Presseau et al., 2013), right sTN (Stankewitz et al., 2010), LC (Keren et al., 2009)). Outside a priori regions of interest FWE correction was performed for all voxels.

521

50.3 (95% CI)) and rather neutral (rose odor: mean = − 4.3 (− 6 to −2.5 (95% CI)), checkerboard: −2.8 (−4.4 to −1.1)). Nociceptive processing We detected significant activations in various brainstem and diencephalic areas including key sites of ascending pain pathways such as the sTN, the VPM of the thalamus and SII, as well as in a pain modulating network including the LC, the CNF, the PAG, the SN, the hypothalamus and the VPM of the thalamus as well as in the insular cortex and the amygdala following nociceptive stimulation of the first and second trigeminal branch (one-sample t test of the pain contrast (ammonia N air)). For a t-score map of the linear contrast ammonia N air see Fig. 2. See Table 1 for a list of significant activations and peak voxels. Contrasting ammonia stimulation over other conditions (visual stimulation, odor stimulation) yielded similar results.

Behavioral data

Parametric modulation To test for the linear effect of pain intensity on areas of the brainstem, we included pain intensity ratings as a parametric modulator of the ammonia stimulus in a separate analysis. On the level of the brainstem, a significant linear correlation with pain intensity ratings was observed for the ipsilateral cuneiform nucleus and the contralateral insula. Fig. 3A shows a t-score map of the linear effect of pain. Table 2 offers a list of significant activations and peak voxels.

Ammonia stimuli were generally rated as quite painful (mean = 70.5 (69.9–75.1)) and unpleasant (mean: 30.5 (28.7–32.4 (95% CI))), whereas olfactory and visual stimuli were generally perceived as less intense (rose odor: 41.5 (39.1–43.8 (95% CI)), checkerboard: 48.4 (46.4–

Psychophysiological interaction analysis As one of our main interests lay in determining functional connectivity between brainstem areas following trigeminal nociceptive stimulation, we conducted a psychophysiological interaction analysis (PPI)

Results

Fig. 2. Brainstem activations following trigeminal nociceptive stimulation. A: Brainstem activations in response to painful ammonia insufflation to the left nostril and anatomical overview of brainstem areas involved in trigeminal nociceptive stimulation: shown is the pain contrast ammonia N air, n = 21, visualization threshold: p b 0.0003, voxel extent = 10. We detected significant activations in the sTN, CNF, PAG, HT and LC. Activation of the sTN survived a whole brain FWE-correction for p b 0.05. B: Brainstem activations found in a previous whole brain fMRI study of trigeminal nociception (Stankewitz et al., 2010: activity of the sTN, n = 20, p b 0.001, uncorr.).

522

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

Table 1 Activations of the brain and brainstem following trigeminal nociceptive stimulation. x, y and z coordinates and T-values are reported for the peak voxel of the cluster in question. Region

Coordinates of the peak voxel (x, y, z)

T-value

sTN ipsilateral sTN contralateral Locus coeruleus contralateral Cuneiform nucleus ipsilateral PAG Substantia nigra contralateral Hypothalamus ipsilateral Amygdala ipsilateral Amygdala contralateral Insular cortex ipsilateral Insular cortex contralateral Thalamus contralateral Thalamus ipsilateral SII ipsilateral SII contralateral

−4 4 8 −11 −5 9 −7 −23 21 −39 39 13 −12 −57 53

8.33⁎⁎ 4.36⁎ 5.08⁎ 6.02⁎ 5.58⁎ 7.58⁎⁎ 5.52⁎ 5.76⁎ 6.32⁎ 6.16⁎ 7.06⁎ 7.43⁎⁎ 7.00⁎ 4.10⁎ 10.03⁎⁎

−45 −42 −39 −25 −28 −15 −15 −4 −2 −18 −1 −14 −18 −13 2

−53 −53 −33 −9 −3 −7 −6 −11 −11 −3 −7 7 6 9 6

⁎ p b 0.05, FWE-corrected (small volume). ⁎⁎ p b 0.05, FWE-corrected (whole volume).

with pain (ammonia N air) as a psychological factor and the left sTN as a seed region. This analysis revealed pain-specific functional connectivity of the left sTN with the contralateral sTN and HT (see Fig. 4 for a t-score map). As we identified the CNF as showing graded activity corresponding to pain intensity ratings, we also conducted a PPI analysis using this region as the seed. The CNF showed enhanced functional connectivity to the RVM during pain. See Fig. 3B for a t-score map of the PPI results

Table 2 Parametric modulation with pain intensity ratings as parametric modulator revealed activity within the ipsilateral cuneiform nucleus and the contralateral insular cortex to correlate with perceived pain intensity. x, y and z coordinates and T-values are reported for the peak voxel of the cluster in question. Region

Coordinates of the peak voxel (x, y, z)

T-value

Cuneiform nucleus ipsilateral Insular cortex contralateral

−6 37

4.19⁎ 5.46⁎

−27 −20

−12 5

⁎ p b 0.05, FWE-corrected (small volume).

from the left CNF and Table 3 for a list of significant activations and peak voxels. Olfactory and visual processing Following olfactory stimulation with rose odor, we found bilateral activations of the amygdala and the insular cortex. Checkerboard stimulation led to strong bilateral activations of the lateral geniculate body as well as the ventral parts of the visual cortex present in our field of view. Discussion Employing an adjusted protocol for high resolution functional brainstem imaging in combination with the two noise correction techniques, we found strong activations in the major conduction sides of trigeminal pain processing such as the caudal part of the ipsilateral sTN, the VPM, bilateral insula, and SII following trigeminal nociceptive processing. As our imaging technique focused on the brainstem, higher

Fig. 3. A: Parametric modulation: of the brainstem centers involved in trigeminal nociceptive processing, the ipsilateral cuneiform nucleus showed a linear activity increase with pain intensity ratings (visualization threshold p b 0.005, voxel extent = 10). B: PPI of the left CNF with pain (ammonia N air) as a psychological factor revealed enhanced functional connectivity of left CNF to the RVM under the influence of pain (visualization threshold p b 0.005, voxel extent = 10).

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

523

Fig. 4. PPI of the left sTN revealed enhanced functional connectivity to the contralateral upper part of the right sTN (bottom right), to the right posterior HT (top right) and to the right thalamus (not shown) under the influence of pain (visualization threshold p b 0.001, voxel extent = 10).

cortical areas such as SI and ACC were located outside of our field of view. An expected activation of the primary sensory cortex could therefore not be detected, but has been shown by our group in earlier studies using a similar set-up and whole brain imaging (Stankewitz et al., 2010). However, as expected, we found a network of various pain modulating centers to be active including the PAG, the LC, CNF, SN and the HT with a much better spatial resolution as compared to the above mentioned previous studies. The observed activations in various brainstem centers following trigeminal nociception are in line with previous animal and anatomical studies which have been able to characterize the fiber connections and anatomy of the trigeminal nociceptive pathway and the so-called descending pain modulating system (Basbaum et al., 2009; Burstein et al., 1998; Edvinsson, 2011; Keren et al., 2009; Millan, 2002; Noseda et al., 2008; Noseda and Burstein, 2013; Usunoff et al., 1997). Although previous functional imaging studies investigating trigeminal nociception have described either activations in the brainstem, such as the dorsal medulla oblongata (Borsook et al., 2006; DaSilva et al., 2002; Mainero et al., 2007; Stankewitz et al., 2010), or the ventral pons (Stankewitz et al., 2010) and the pontomesencephalic junction (Hummel et al., 2005), the detected activations have, due to the whole brain approach, generally been rather large with a poor spatial resolution. A clear distinction between certain brainstem nuclei was consequently Table 3 Functional connectivity of the left spinal trigeminal nucleus and the left CNF under the influence of pain. x, y and z coordinates and T-values are reported for the peak voxel of the cluster in question. Seed region: left sTN Region

Coordinates of the peak voxel (x, y, z)

sTN contralateral Hypothalamus contralateral

4 5

−40 −21

−43 −3

T-value 5.11⁎ 4.39⁎

Seed region: left CNF Region

Coordinates of the peak voxel (x, y, z)

RVM

2

⁎ p b 0.05, FWE-corrected (small volume).

−38

−46

T-value 3.99⁎

impossible. By using the described brainstem optimized imaging and processing protocol with a high spatial resolution and two different noise correction techniques for high resolution brainstem fMRI, we were able to achieve a more sophisticated view of brainstem areas involved in trigeminonociceptive processing. This allowed a clear distinction between certain brainstem areas and an assignment of the found activations to certain cell groups within the brainstem (see Fig. 2). However, although the higher spatial resolution makes a clearer separation of certain brainstem nuclei possible, it might still not be high enough to separate activations in areas with a very close spatial relationship of certain brainstem centers or between certain subnuclei of an area. Also, as another limitation, other noise correction techniques such as RETROICOR analyses have been shown to outperform the selective averaging method used in this study regarding functional imaging of the spinal cord (Kong et al., 2012). However, the selective averaging method of Deckers (Deckers et al., 2006) as applied in this study has previously been shown to be at least equivalent to RETROICOR. Additionally, other noise correction techniques such as RETROICOR or ICA-based approaches are also well established and broadly accepted (Brooks et al., 2013; Kong et al., 2012) but work with much more a priori assumptions than the Deckers method. We thus chose this also well established noise correction technique in the current study. Notably, although nociceptive stimuli were applied unilaterally, we detected bilateral activations already at the level of the spinal trigeminal nuclei. Also, we found enhanced functional connectivity of the left and the right sTN, thus suggesting direct fiber connections between the caudal (pain receiving) part of the left sTN and the upper part of the right sTN. Indeed neuronal labeling experiments in animals have identified fiber connections between both sTN (Jacquin et al., 1990a; Usunoff et al., 1997). Whether the bilateral sTN activation in our experiments and the pain-dependent functional connectivity between both nuclei are due to direct co-activation of the contralateral trigeminal nucleus or rather an activation of inhibitory interneurons close to, or within the contralateral sTN, is not evident from our data. However, as the thalamus and SII were activated also ipsilateral to the stimulation side, a direct co-activation of the contralateral sTN seems to be more likely. Another possible explanation is a direct innervation of the contralateral spinal trigeminal nucleus by trigeminal primary afferents of the

524

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525

ophthalmic branch, as has been shown in previous animal studies (Jacquin et al., 1990b; Marfurt and Rajchert, 1991; Panneton et al., 1991; Usunoff et al., 1997). The physiological function of these fiber connections has not yet been satisfactorily explained. Again, as the neuronal correlate for the observed contralateral sTN-activation can be due to activity of second order trigeminal neurons as well as to activation of inhibitory interneurons located close to the right sTN, we can currently not make a deduction about the physiological function of these crossing fibers. Theoretically it is possible, that gaseous ammonia applied to the left nostril might spread and coactivate peripheral c fibers of the right trigeminal nerve as well. However, as pain perception in our experiments is strictly left sided and subjects were advised not to breathe through the nose during the experiment, this explanation is not likely. Another interesting finding is the positive correlation of intensity ratings for painful stimuli and activity of the ipsilateral CNF. The CNF is part of the mesencephalic reticular formation. It is thus supposed to be a potent downstream mediator of pain (Carlson et al., 2004, 2005; Edvinsson, 2011; Rempe et al., 2014) and is involved in central sensitization mechanisms (Zambreanu et al., 2005) as well as “opioid induced hyperalgesia” (Wanigasekera et al., 2011). Consequently, our data can be interpreted in two possible ways: an increased activity of the CNF – possibly as a consequence of central sensitization processes – might lead to a more intense pain perception and thus corroborate a pain facilitating function of this area of the brainstem. Or, the increased activity of the CNF could be a sign of enhanced activity of antinociceptive pathways following more painful stimuli, thus supporting a pain inhibitory function of the CNF. As however the ammonia concentration applied to the participant's nostril is rather constant over the experiment and the position of the tube is not altered, the first explanation is more likely. Additionally, we were able to show enhanced connectivity of the left CNF to the RVM under the influence of pain. The CNF is, together with the PAG, the major region of afferent input to the RVM (Abols and Basbaum, 1981; Behbehani and Zemlan, 1986). CNF and RVM both can – due to their histological characteristics – either facilitate or inhibit pain perception (Fields et al., 1983; Haws et al., 1989). The observed functional connectivity between both regions might thus be an important mechanism in the modulation of central pain perception. Some of the areas we found to be activated during pain have previously been implied to play a certain role in headache pathophysiology: especially the hypothalamus has been shown to be specifically active during the pain state of acute cluster headache attacks as opposed to the painfree interval (May et al., 1998; Morelli et al., 2013) and to play a certain role in spontaneous migraine attacks (Denuelle et al., 2007). We observed a significant activation of the hypothalamic area ipsilateral to the stimulation site and enhanced functional connectivity of the left sTN to the contralateral HT, thus implying a physiological role of the hypothalamus in trigeminal pain processing per se. In migraine, structures of the ventral and dorsal mesencephalon have been shown to be activated during attacks (Cao et al., 2002; Denuelle et al., 2007). The experimental approach to investigate trigeminal pain processing in this study with a standardized trigeminal stimulation, an adapted imaging protocol and retrospective noise correction approaches might thus lead to a better understanding of the physiological mechanisms of trigeminal pain processing. As our paradigm also provided reliable activations in areas known to be activated during olfactory and visual processing (Stankewitz and May, 2011), this makes it best suited to study the pathophysiological mechanisms of certain headache disorders involving photo- and osmophobia. The investigation of these disorders using the proposed technique and paradigm can offer a unique opportunity to further elucidate headache pathophysiology. Taken together, by using an optimized approach for high resolution brainstem echoplanar imaging, a smaller field of view and a combination of two noise correction techniques we were able to achieve a much higher spatial resolution and stronger activations in multiple regions of the human brainstem involved in trigeminonociceptive

processing. Also on the level of the brainstem we identified the CNF as being specifically related to perceived pain intensity and showed enhanced functional connectivity of this region to the RVM. Additionally, pain-related functional connectivity of the caudal part of the left sTN to the rostral part of the right sTN and the posterior hypothalamic gray area was observed. This allowed a clear distinction between certain brainstem areas and an assignment of the found activations to certain cell groups within the brainstem as well as functional connectivity analyses between different brainstem nuclei. Acknowledgment The authors thank Dr. Jürgen Finsterbusch, head of the MR-Physics Group, Institute for Systems Neuroscience, University Medical Center Eppendorf, Hamburg, Germany, for optimizing the employed MRIacquisition protocol. Funding sources This work was supported by the 7th Framework EU-Project EuroHeadPain (#602633) to A.M. References Abols, I.A., Basbaum, A.I., 1981. Afferent connections of the rostral medulla of the cat: a neural substrate for midbrain–medullary interactions in the modulation of pain. J. Comp. Neurol. 201, 285–297. http://dx.doi.org/10.1002/cne.902010211. Afridi, S.K., Giffin, N.J., Kaube, H., Friston, K.J., Ward, N.S., Frackowiak, R.S.J., Goadsby, P.J., 2005. A positron emission tomographic study in spontaneous migraine. Arch. Neurol. 62, 1270–1275. http://dx.doi.org/10.1001/archneur.62.8.1270. Ashburner, J., Friston, K.J., 2005. Unified segmentation. NeuroImage 26, 839–851. http:// dx.doi.org/10.1016/j.neuroimage.2005.02.018. Bahra, A., Matharu, M.S., Buchel, C., Frackowiak, R.S., Goadsby, P.J., 2001. Brainstem activation specific to migraine headache. Lancet 357, 1016–1017. Basbaum, A.I., Bautista, D.M., Scherrer, G., Julius, D., 2009. Cellular and molecular mechanisms of pain. Cell 139, 267–284. http://dx.doi.org/10.1016/j.cell.2009.09.028. Behbehani, M.M., Zemlan, F.P., 1986. Response of nucleus raphe magnus neurons to electrical stimulation of nucleus cuneiformis: role of acetylcholine. Brain Res. 369, 110–118. Beissner, F., 2015. Functional MRI of the brainstem: common problems and their solutions. Clin. Neuroradiol. http://dx.doi.org/10.1007/s00062-015-0404-0. Beissner, F., Deichmann, R., Baudrexel, S., 2011. fMRI of the brainstem using dual-echo EPI. NeuroImage 55, 1593–1599. http://dx.doi.org/10.1016/j.neuroimage.2011.01.042. Borsook, D., Burstein, R., Moulton, E., Becerra, L., 2006. Functional imaging of the trigeminal system: applications to migraine pathophysiology. Headache 46 (Suppl. 1), S32–S38. Boyle, J.A., Heinke, M., Gerber, J., Frasnelli, J., Hummel, T., 2007. Cerebral activation to intranasal chemosensory trigeminal stimulation. Chem. Senses 32, 343–353. http://dx. doi.org/10.1093/chemse/bjm004. Brooks, J.C.W., Faull, O.K., Pattinson, K.T.S., Jenkinson, M., 2013. Physiological noise in brainstem FMRI. Front. Hum. Neurosci. 7, 623. http://dx.doi.org/10.3389/fnhum. 2013.00623. Burstein, R., Yamamura, H., Malick, A., Strassman, A.M., 1998. Chemical stimulation of the intracranial dura induces enhanced responses to facial stimulation in brain stem trigeminal neurons. J. Neurophysiol. 79, 964–982. Cao, Y., Aurora, S.K., Nagesh, V., Patel, S.C., Welch, K.M.A., 2002. Functional MRI-BOLD of brainstem structures during visually triggered migraine. Neurology 59, 72–78. Capra, N.F., Dessem, D., 1992. Central connections of trigeminal primary afferent neurons: topographical and functional considerations. Crit. Rev. Oral Biol. Med. 4, 1–52. http:// dx.doi.org/10.1177/10454411920040010101. Carlson, J.D., Iacono, R.P., Maeda, G., 2004. Nociceptive excited and inhibited neurons within the pedunculopontine tegmental nucleus and cuneiform nucleus. Brain Res. 1013, 182–187. http://dx.doi.org/10.1016/j.brainres.2004.03.069. Carlson, J.D., Selden, N.R., Heinricher, M.M., 2005. Nocifensive reflex-related on- and offcells in the pedunculopontine tegmental nucleus, cuneiform nucleus, and lateral dorsal tegmental nucleus. Brain Res. 1063, 187–194. http://dx.doi.org/10.1016/j. brainres.2005.09.036. Carpenter, M.B., Hanna, G.R., 1961. Fiber projections from the spinal trigeminal nucleus in the cat. J. Comp. Neurol. 117, 117–131. Coupé, P., Yger, P., Barillot, C., 2006. Fast non local means denoising for 3D MR images. Med. Image Comput. Comput.-Assist. Interv. 9, pp. 33–40 DaSilva, A.F.M., Becerra, L., Makris, N., Strassman, A.M., Gonzalez, R.G., Geatrakis, N., Borsook, D., 2002. Somatotopic activation in the human trigeminal pain pathway. J. Neurosci. 22, 8183–8192. Deckers, R.H.R., van Gelderen, P., Ries, M., Barret, O., Duyn, J.H., Ikonomidou, V.N., Fukunaga, M., Glover, G.H., de Zwart, J.A., 2006. An adaptive filter for suppression of cardiac and respiratory noise in MRI time series data. NeuroImage 33, 1072–1081. http://dx.doi.org/10.1016/j.neuroimage.2006.08.006.

L.H. Schulte et al. / NeuroImage 124 (2016) 518–525 Denuelle, M., Fabre, N., Payoux, P., Chollet, F., Geraud, G., 2007. Hypothalamic activation in spontaneous migraine attacks. Headache 47, 1418–1426. http://dx.doi.org/10.1111/j. 1526-4610.2007.00776.x. Edvinsson, L., 2011. Tracing neural connections to pain pathways with relevance to primary headaches. Cephalalgia Int. J. Headache 31, 737–747. http://dx.doi.org/10. 1177/0333102411398152. Fields, H.L., Bry, J., Hentall, I., Zorman, G., 1983. The activity of neurons in the rostral medulla of the rat during withdrawal from noxious heat. J. Neurosci. Off. J. Soc. Neurosci. 3, 2545–2552. Friston, K.J., Buechel, C., Fink, G.R., Morris, J., Rolls, E., Dolan, R.J., 1997. Psychophysiological and modulatory interactions in neuroimaging. NeuroImage 6, 218–229. http://dx.doi. org/10.1006/nimg.1997.0291. Harvey, A.K., Pattinson, K.T.S., Brooks, J.C.W., Mayhew, S.D., Jenkinson, M., Wise, R.G., 2008. Brainstem functional magnetic resonance imaging: disentangling signal from physiological noise. J. Magn. Reson. Imaging 28, 1337–1344. http://dx.doi.org/10. 1002/jmri.21623. Haws, C.M., Williamson, A.M., Fields, H.L., 1989. Putative nociceptive modulatory neurons in the dorsolateral pontomesencephalic reticular formation. Brain Res. 483, 272–282. Hayashi, H., 1985a. Morphology of central terminations of intra-axonally stained, large, myelinated primary afferent fibers from facial skin in the rat. J. Comp. Neurol. 237, 195–215. http://dx.doi.org/10.1002/cne.902370205. Hayashi, H., 1985b. Morphology of terminations of small and large myelinated trigeminal primary afferent fibers in the cat. J. Comp. Neurol. 240, 71–89. http://dx.doi.org/10. 1002/cne.902400106. Hummel, T., Doty, R.L., Yousem, D.M., 2005. Functional MRI of intranasal chemosensory trigeminal activation. Chem. Senses 30, i205–i206. http://dx.doi.org/10.1093/chemse/ bjh186. Iannilli, E., Del Gratta, C., Gerber, J.C., Romani, G.L., Hummel, T., 2008. Trigeminal activation using chemical, electrical, and mechanical stimuli. Pain 139, 376–388. http:// dx.doi.org/10.1016/j.pain.2008.05.007. Jacquin, M.F., Chiaia, N.L., Haring, J.H., Rhoades, R.W., 1990a. Intersubnuclear connections within the rat trigeminal brainstem complex. Somatosens. Mot. Res. 7, 399–420. Jacquin, M.F., Chiaia, N.L., Rhoades, R.W., 1990b. Trigeminal projections to contralateral dorsal horn: central extent, peripheral origins, and plasticity. Somatosens. Mot. Res. 7, 153–183. Keren, N.I., Lozar, C.T., Harris, K.C., Morgan, P.S., Eckert, M.A., 2009. In vivo mapping of the human locus coeruleus. NeuroImage 47, 1261–1267. http://dx.doi.org/10.1016/j. neuroimage.2009.06.012. Kong, Y., Jenkinson, M., Andersson, J., Tracey, I., Brooks, J.C.W., 2012. Assessment of physiological noise modelling methods for functional imaging of the spinal cord. NeuroImage 60, 1538–1549. http://dx.doi.org/10.1016/j.neuroimage.2011.11.077. Mainero, C., Zhang, W.-T., Kumar, A., Rosen, B.R., Sorensen, A.G., 2007. Mapping the spinal and supraspinal pathways of dynamic mechanical allodynia in the human trigeminal system using cardiac-gated fMRI. NeuroImage 35, 1201–1210. http://dx.doi.org/10. 1016/j.neuroimage.2007.01.024. Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. NeuroImage 19, 1233–1239. Marfurt, C.F., Rajchert, D.M., 1991. Trigeminal primary afferent projections to “non-trigeminal” areas of the rat central nervous system. J. Comp. Neurol. 303, 489–511. http://dx.doi.org/10.1002/cne.903030313. May, A., Bahra, A., Büchel, C., Frackowiak, R.S., Goadsby, P.J., 1998. Hypothalamic activation in cluster headache attacks. Lancet 352, 275–278. http://dx.doi.org/10.1016/ S0140-6736(98)02470-2. McCulloch, P.F., DiNovo, K.M., Westerhaus, D.J., Vizinas, T.A., Peevey, J.F., Lach, M.A., Czarnocki, P., 2013. Trigeminal medullary dorsal horn neurons activated by nasal stimulation coexpress AMPA, NMDA, and NK1 receptors. ISRN Neurosci. 2013, 152567. http://dx.doi.org/10.1155/2013/152567.

525

Millan, M.J., 2002. Descending control of pain. Prog. Neurobiol. 66, 355–474. Morelli, N., Rota, E., Gori, S., Guidetti, D., Michieletti, E., Simone, R.D., Salle, F.D., 2013. Brainstem activation in cluster headache: an adaptive behavioural response? Cephalalgia 33, 416–420. http://dx.doi.org/10.1177/0333102412474505. Noseda, R., Burstein, R., 2013. Migraine pathophysiology: anatomy of the trigeminovascular pathway and associated neurological symptoms, CSD, sensitization and modulation of pain. Pain 154 (Suppl. 1). http://dx.doi.org/10. 1016/j.pain.2013.07.021. Noseda, R., Monconduit, L., Constandil, L., Chalus, M., Villanueva, L., 2008. Central nervous system networks involved in the processing of meningeal and cutaneous inputs from the ophthalmic branch of the trigeminal nerve in the rat. Cephalalgia 28, 813–824. http://dx.doi.org/10.1111/j.1468-2982.2008.01588.x. Panneton, W.M., Klein, B.G., Jacquin, M.F., 1991. Trigeminal projections to contralateral dorsal horn originate in midline hairy skin. Somatosens. Mot. Res. 8, 165–173. Rempe, T., Wolff, S., Riedel, C., Baron, R., Stroman, P.W., Jansen, O., Gierthmühlen, J., 2014. Spinal and supraspinal processing of thermal stimuli: an fMRI study: processing of thermal stimuli. J. Magn. Reson. Imaging. http://dx.doi.org/10.1002/ jmri.24627 (n/a–n/a). Stankewitz, A., May, A., 2011. Increased limbic and brainstem activity during migraine attacks following olfactory stimulation. Neurology 77, 476–482. http://dx.doi.org/10. 1212/WNL.0b013e318227e4a8. Stankewitz, A., Voit, H.L., Bingel, U., Peschke, C., May, A., 2010. A new trigeminonociceptive stimulation model for event-related fMRI. Cephalalgia Int. J. Headache 30, 475–485. http://dx.doi.org/10.1111/j.1468-2982.2009.01968.x. Stankewitz, A., Aderjan, D., Eippert, F., May, A., 2011. Trigeminal nociceptive transmission in migraineurs predicts migraine attacks. J. Neurosci. Off. J. Soc. Neurosci. 31, 1937–1943. http://dx.doi.org/10.1523/JNEUROSCI.4496-10.2011. Stewart, W.A., King, R.B., 1963. Fiber projections from the nucleus caudalis of the spinal trigeminal nucleus. J. Comp. Neurol. 121, 271–286. Tiwari, R.K., King, R.B., 1974. Fiber projections from trigeminal nucleus caudalis in primate (squirrel monkey and baboon). J. Comp. Neurol. 158, 191–205. http://dx.doi.org/10. 1002/cne.901580206. Usunoff, K.G., Marani, E., Schoen, J.H., 1997. The trigeminal system in man. Adv. Anat. Embryol. Cell Biol. 136, 1–126 (I–X). Vachon-Presseau, E., Martel, M.-O., Roy, M., Caron, E., Albouy, G., Marin, M.-F., Plante, I., Sullivan, M.J., Lupien, S.J., Rainville, P., 2013. Acute stress contributes to individual differences in pain and pain-related brain activity in healthy and chronic pain patients. J. Neurosci. Off. J. Soc. Neurosci. 33, 6826–6833. http://dx.doi.org/10.1523/JNEUROSCI. 4584-12.2013. Wallenberg, A., 1900. Sekundäre sensible Bahnen im Gehirnstamme des Kaninchens, ihre gegenseitige Lage und ihre Bedeutung für den Aufbau des Thalamus. Anat. Anz. 12, 95–110. Wanigasekera, V., Lee, M.C.H., Rogers, R., Hu, P., Tracey, I., 2011. Neural correlates of an injury-free model of central sensitization induced by opioid withdrawal in humans. J. Neurosci. Off. J. Soc. Neurosci. 31, 2835–2842. http://dx.doi.org/10. 1523/JNEUROSCI.5412-10.2011. Weiller, C., May, A., Limmroth, V., Jüptner, M., Kaube, H., Schayck, R.V., Coenen, H.H., Diener, H.C., 1995. Brain stem activation in spontaneous human migraine attacks. Nat. Med. 1, 658–660. Zambreanu, L., Wise, R.G., Brooks, J.C.W., Iannetti, G.D., Tracey, I., 2005. A role for the brainstem in central sensitisation in humans. Evidence from functional magnetic resonance imaging. Pain 114, 397–407. http://dx.doi.org/10.1016/j.pain.2005.01.005. Zwergal, A., la Fougère, C., Lorenzl, S., Rominger, A., Xiong, G., Deutschenbaur, L., Schöberl, F., Linn, J., Dieterich, M., Brandt, T., Strupp, M., Bartenstein, P., Jahn, K., 2013. Functional disturbance of the locomotor network in progressive supranuclear palsy. Neurology 80, 634–641. http://dx.doi.org/10.1212/WNL.0b013e318281cc43.