|Home | About | Journals | Submit | Contact Us | Français|
Previous studies suggest that nocebo effects, sometimes termed “negative placebo effects,” can contribute appreciably to a variety of medical symptoms and adverse events in clinical trials and medical care. In this study, using a within-subject design, we combined fMRI and an expectation / conditioning manipulation model to investigate the neural substrates of nocebo hyperalgesia using heat pain on the right forearm. Thirteen subjects completed the study. Results showed that after administering inert treatment, subjective pain intensity ratings increased significantly more on nocebo regions as compared with the control regions where no expectancy / conditioning manipulation was performed. fMRI analysis of hyperalgesic nocebo responses to identical calibrated noxious stimuli showed signal increases in brain regions including bilateral dorsal ACC, insula, superior temporal gyrus; left frontal and parietal operculum, medial frontal gyrus, orbital prefrontal cortex, superior parietal lobule and hippocampus; right claustrum / putamen, lateral prefrontal gyrus and middle temporal gyrus. Functional connectivity analysis of spontaneous resting-state fMRI data from the same cohort of subjects showed a correlation between two seed regions (frontal operculum and left hippocampus) pain network including bilateral insula, operculum, ACC, and left S1 / M1. In conclusion, we found evidence that nocebo hyperalgesia may be predominantly produced through an affective-cognitive pain pathway (medial pain system) and the left hippocampus may play an important role in this process.
Placebo effects have received much attention in recent decades, and with the advent of brain imaging tools, our understanding of their neurobiology has greatly expanded (Benedetti et al., 2005; Colloca and Benedetti, 2005; Benedetti, 2007; Kong et al., 2007). In contrast, nocebo effects – adverse effects or a worsening of symptoms on account of expectation or suggestion regarding an inert treatment – have received relatively scant attention from neuroscience.
Previous studies suggest that nocebo effects, sometimes termed “negative placebo effects,” contribute appreciably to a variety of medical symptoms (Barsky and Borus, 1999; Barsky et al., 2002), adverse events in clinical trials and medical care (Myers et al., 1987; Roscoe et al., 2000; Reuter et al., 2003; Kaptchuk et al., 2006), and public health “mass psychogenic illness” outbreaks (Jones et al., 2000). For instance, Ko and colleagues (Ko et al., 2003) found patients receiving a beta-blocker and patients receiving placebo report comparable levels of common side effects, including depressive symptoms, fatigue, and sexual dysfunction.
In one of the few nocebo mechanism studies to date, Benedetti and colleagues (Benedetti et al., 1997) reported that while proglumide, a nonspecific cholecystokinin (CCK) antagonist, could counteract nocebo-induced hyperalgesia, the opioid antagonist naloxone had no effect on nocebo responses. In a following study, Bendetti and colleagues (Benedetti et al., 2006) found that while the benzodiazepine diazepam could block both nocebo hyperalagesia and hypothalamic-pituitary-adrenal (HPA) hyperactivity, proglumide could block only the former. As such, benzodiazepines may act on nocebo-induced anxiety, whereas proglumide may act specifically on the CCK-mediated link between pain and anxiety. An animal study (Andre et al., 2005) further documented CCK-B receptor antagonist CI-988’s ability to block anxiety-associated hyperalgesia, suggesting hyperalgesic effects may represent an emergent property of the mechanisms linking anxiety to pain.
Previous studies have suggested that the central pain matrix consists of roughly two parallel subsystems, the lateral (sensory-discriminatory) and medial (affective-cognitive evaluation) system (Apkarian et al., 2005; Tracey and Mantyh, 2007). The lateral system projects through the lateral thalamic nuclei to the cortex and includes the primary and secondary somatosensory cortices. The medial system projects through the medial thalamic nuclei to the cortex and includes the anterior cingulate cortex (ACC), insula and prefrontal cortices (Price, 2000; Rainville, 2002; Vogt, 2005).
In this current experiment, we used fMRI and a modified expectancy / conditioning manipulation model (Montgomery and Kirsch, 1997; Price et al., 1999; Wager et al., 2004; Kong et al., 2006b) to investigate the neural substrates of hyperalgesic nocebo. At the very beginning of fMRI scanning session, we also collected a six-minute spontaneous fMRI data set to investigate functional connectivity (Raichle and Mintun, 2006; Buckner and Vincent, 2007; Fox and Raichle, 2007) among the brain regions observed to be activated in the subsequent analysis of the nocebo scans. Based on previous nocebo behavioral studies (Benedetti et al., 1997; Andre et al., 2005; Benedetti et al., 2006) indicating a role for anxiety in nocebo, we hypothesize that nocebo hyperalgesia may mainly arise from increased activation of the medial pain matrix.
Twenty healthy, right-handed subjects, cleared for fMRI participation, enrolled in this study. To conceal the nocebo aim of the experiment, we told subjects we would be studying acupuncture’s effects on the brain. Because we wanted our intervention to be novel for patients, we employed a validated sham acupuncture device, and allowed only acupuncture-naïve subjects to participate. Experiments were conducted with written consent of each subject and approved by the Massachusetts General Hospital Institutional Review Board. All subjects were debriefed at the end of the experiment as to the true nature of the experiment.
Subjects participated in two behavioral testing sessions and one fMRI scanning session. Each session was separated by a minimum of three days.
Calibrated thermal pain stimuli were delivered to the right medial forearm using a TSA-2001 Thermal Sensory Analyzer with a 3 cm ×?3 cm probe (Medoc Advanced Medical Systems, Rimat Yishai, Israel) running proprietary computerized visual analog scale software (COVAS). Each stimulus was initiated from a 32°C baseline, increased to a target temperature (45.5 ~ 51°C), and presented for 12 seconds, including a 2.5 second ramp up and ramp down. Inter-stimulus intervals ranged from 24–30 seconds.
Gracely Sensory and Affective scales (Gracely et al., 1978a; Gracely et al., 1978b) were used to measure subjective pain ratings. To ensure consistent pain administration, a 2×3 grid was drawn in marker along the palmar side of the forearm, with three boxes each on radial and ulnar sides. We placed the thermal probe in one box of the grid for each stimulus sequence (Figure 1).
We used the first behavioral session to familiarize subjects with the rating scales and determine appropriate stimulus intensities using methods employed in our previous studies (Kong et al., 2005; Kong et al., 2006b; Kong et al., 2006a). Temperatures eliciting subjective intensity ratings in the LOW pain range (~ 5; weak on the 0–20 Sensory Scale), MID pain range (~ 9; mild), and HIGH pain range (~ 15; strong) were selected for each individual and used in subsequent sessions. Next we applied series of 8 Random Pain noxious stimuli (4 HIGH and 4 LOW applied in random order, indicated by the abbreviation RP) and series of 6 Identical Pain noxious stimuli (6 identical MID, indicated by IP) to the right arm. Temperatures were adjusted when necessary to ensure that each subject’s subjective ratings of HIGH, LOW, and MID remained in the desired range, as these would be used in the following sessions.
We used Session 2 to manipulate subjects’ expectancy to sham acupuncture treatment using a method modified from our previous work (Montgomery and Kirsch, 1997; Price et al., 1999; Wager et al., 2004; Kong et al., 2006b).
At the beginning of Session 2, subjects read a script stating: 1) responses to acupuncture can be positive, neutral, or negative and 2) a given subject’s response tends to remain consistent across sessions. Subjects then viewed a Traditional Chinese Medicine meridian diagram and were told: 3) according previous literature, acupuncture would only produce effects (positive or negative) on the side of the arm where the meridian passed through but not on the other side of the arm. To balance the design, half the subjects were then shown accurate diagrams (real diagram) of the Large Intestine (LI) Meridian passing through the radial side of the arm, while the other half viewed a modified diagram (fake diagram), showing the LI Meridian passing through the ulnar side of the arm. (For clarity, we refer to meridian (real or fake) and non-meridian (real or fake) sites as nocebo and control sites respectively).
Next the same RP were administered to the bottom two boxes of the 2×3 grid along the arm and MID pain IP were applied to the top four (Figure 1A).
To proceed in the study, subjects had to consistently rate HIGH pain greater than LOW pain, and report approximately equivalent ratings (less than a 1.5 on average intensity rating difference) to MID pain on the radial and ulnar sides of their arm.
Sham electro-acupuncture was then performed at two acupoints on the right hand (LI 4 and LI 3) using the validated, non-penetrating Streitberger sham acupuncture device (Streitberger and Kleinhenz, 1998; Kleinhenz et al., 1999; Kong et al., 2006b; McManus et al., 2007). As in our previous methods (Kong et al., 2005), the sham needles were connected to an electro-acupuncture device (OMS Medical Supplies IC-1107), with no current applied.
After treatment, we told subjects they would be receiving the same stimuli series administered before treatment. In actuality, on the nocebo side of the arm, we used an increased IP stimuli (iIP, HIGH instead of MID intensities), giving subjects an unmistakable experience of hypersensitivity. On the control side, IP were maintained at pre-treatment MID levels (Figure 1A).
In this session, subjects also completed an expectancy rating form (−10 = extreme pain sensitivity, to 0 = neutral no change, to 10 = complete pain relief) once before sham treatment and once after the expectancy manipulation procedure.
Session 3 was performed in the fMRI scanner. Subjects were told we would be repeating Session 2 procedures. In actuality, while most procedures matched those performed in Session 2, the exception was that after treatment, only one increased IP (iIP) was applied on the nocebo side of the arm to boost subjects’ memory of hyperalgesia, administer an additional series of the conditioning trial, and provide an experience of hyperalgesia closer in time to the nocebo test. Original MID IP stimuli were applied on all other regions of the arm. To minimize the scan time, only pain sensory rating data were collected in this session. The differences between pre- and post-treatment pain ratings and brain activation during these final four sequences (two MID IP on nocebo and control sides each) were the primary outcomes of this study (Figure 1B).
At the beginning of Session 3 and after the expectancy boost, subjects were required to complete the expectancy rating forms again.
Brain imaging was performed with a 3-axis gradient head coil in a 3 Tesla whole body Siemens MRI System equipped for echo planar imaging. Thirty axial slices (4 mm thick with 1 mm skip) parallel to the anterior and posterior commissure covering the whole brain were imaged with 2000 ms TR, 40 ms TE, 90° flip angle and 3.13 × 3.13 mm in-plane spatial resolution. A high-resolution 3D MPRAGE sequence was also collected for anatomic localization.
At the beginning of the scanning session, a six-minute resting status scan was performed, during which subjects were instructed to close their eyes and relax. Afterward, fMRI scanning was performed during the administration of pain (RP and IP) before and after treatment using an experimental paradigm similar to our previous studies (Kong et al., 2006b; Kong et al., 2006a). During scanning, subjects were instructed to focus on a small black fixation cross in the center of a screen in front of them. The cross turned red to cue the onset and duration of each stimulus (12s) and then turned black for a variable duration (4, 6, or 8 sec). Then, the Sensory Box Scale was displayed on the screen (8s) and subjects used a button press device controlling a pointer to indicate their subjective ratings (Figure 1c).
Pre-processing and statistical analyses were performed using SPM2 software (Wellcome Department of Cognitive Neurology). Pre-processing included motion correction, normalization to MNI stereotactic space, and spatial smoothing with an 8 mm Gaussian kernel.
For each subject, a general linear model (GLM) design matrix was calculated, including all pain functional runs used to test nocebo effects (4 runs before and after treatment on both nocebo and control sites). Then, calculations were performed on the contrast between pre-treatment HIGH pain and LOW pain on both nocebo and control sites. Finally, the contrast comparing post- minus pre- differences in response to identical IP stimuli on nocebo sites and control sites —nocebo increases relative to control (nocebo (post-pre) – control (post-pre)) were calculated. Low-frequency noise was removed with a high-pass filter applied with default values (128s) to the fMRI time series at each voxel.
Group analysis was performed using a random-effects model. A one-way t-test was performed to determine group activation for each generated contrast as described above. As in previous studies, to elucidate pain-intensity correlated brain regions with which we would later use to test for nocebo effects, we conducted an initial group comparison of all pre-treatment HIGH and LOW pain sequences (Wager et al., 2004; Kong et al., 2006b), constructing a mask for subsequent group analysis (with threshold set at voxel-wise p < 0.005 uncorrected with 20 contiguous voxels). Based on previous studies (Petrovic and Ingvar, 2002; Benedetti et al., 2005) implicating a role for DLPFC and OPFC in pain modulation, these two regions were also included as a priori regions of interest (ROIs). Next, to elucidate the brain regions involved in nocebo effect, the contrast comparing post- minus pre- differences in response to application of IP stimuli on nocebo and control sites was calculated.
In a second level analysis, a simple regression (correlation) analysis between each subject’s fMRI signal changes (nocebo (post-pre) – control (post-pre)) and corresponding subjective pain rating changes was also calculated.
The threshold for pre-defined ROIs (either within the mask or DLPFC and OPFC) was set at voxel-wise p < 0.005 uncorrected with 20 contiguous voxels. A threshold of voxel-wise p < 0.05 corrected for 5 contiguous voxels was used for activation in other regions.
To further investigate the functional correlation between regions, two brain regions observed in the fMRI analysis were used as seed regions for a functional connectivity study on the resting status data collected at the beginning of fMRI scan session. Methods for functional connectivity analysis were adopted from previous studies (Fox et al., 2005; Andrews-Hanna et al., 2007; Vincent et al., 2007).
In summary, functional data were first preprocessed to decrease image artifacts and between-slice timing differences, and to eliminate differences in odd/even slice intensity. Rigid body translation and rotation was used to reduce within and across-run head movement. Data were resampled to 2mm isotropic voxels after transforming anatomical and functional data to atlas space.
The functional connectivity analysis required additional filtering of low- and high-frequency components (0.009Hz < f < 0.083Hz) and spatial 8mm Gaussian kernel smoothing. Other variables that were simultaneously regressed included movement parameters, whole brain signal, lateral ventricle mean signals, deep white matter ROI signal, and the first temporal derivative of each time course. A resulting time course was used in the subsequent analysis.
Then, correlation maps between seed regions (left operculum and left hippocampus) and all voxels across the whole brain were performed. Analysis produced seed region-whole brain voxel correlation coefficients. Fisher’s r-to-z transformation was used to convert correlation maps into z maps. Group effects were tested with a random-effect analyses using one sample t-test. The threshold was set at voxel-wise p<0.001 uncorrected for 20 contiguous voxels.
Thirteen of twenty consenting volunteers completed this study (5 males, mean age 26.3 years ± 3.6 SD). Six subjects were dropped on account of their unreliable HIGH and LOW pain ratings, unbalanced pain perceptions on ulnar and radial sides of the arm, or anxiety regarding scanning prior to beginning Session 3. Fourteen subjects completed the scan session. Data from one scanned subject was excluded on account of excessive head movement during scanning (exceeding 5 mm in two functional runs).
We used Session 3 pre- and post- treatment pain sensory intensity rating differences between nocebo and control sites to detect nocebo hyperalgesia. Pre- and post-ratings for identical pain stimuli (mean ± SD) were calculated at 9.0 ± 2.5 and 11.4 ± 2.1 on nocebo sites, and 8.7 ± 2.8 and 10.1 ± 2.1 on control sites. The non-parametric test (Wilcoxon Signed Ranks test) was used to compare the pre-treatment pain ratings and pain rating changes (post- minus pre-) after treatment between nocebo and control sites. The results indicated that there is no difference in pain ratings between the nocebo and control sites (p = 0.33) before treatment. After treatment, subjective pain rating increases on the nocebo sites were significantly greater than those on the control sites (p = 0.021).
On the Expectation Scale, the subjects’ expectancy ratings (mean ± SD) significantly decreased from initially positive levels before manipulation (2.6 ± 2.7) to negative levels after manipulation (−5.1 ± 2.0) with all subjects reporting a negative expectation of acupuncture’s effect on pain in session 2 (Wilcoxon Signed Rank test, p = 0.002). When entering Session 3, subjects’ expectation ratings were maintained at significantly negative levels compared with positive expectancy ratings before manipulation in session 2. The expectancy ratings were −4.8 ± 2.2 at the beginning of session 3 (p = 0.002) and −5.3 ± 1.8 after the Session 3 “expectancy boost” (p = 0.002).
The comparison between all pre-treatment HIGH and LOW pain stimuli (HIGH pain > LOW pain) applied during the RP sequence before treatment yielded significant activations in the entire predicted network of pain sensitive regions, including bilateral insular / opercular cortices, dorsal anterior cingulate cortex (dACC), superior parietal lobule, superior & middle frontal gyrus, superior & middle temporal gyrus, hippocampus, thalamus, caudate, cerebellum, brain stem (PAG and pons); left (contralateral) SI / M1 corresponding to the arm, medial frontal gyrus, inferior parietal lobule; and right inferior frontal gyrus, parahippocampus, angular gyrus and precuneus (Figure 2A). This result is consistent with previous studies (Wager et al., 2004; Kong et al., 2006b; Kong et al., 2006a) and corresponds to the reported subjective pain ratings (12.9 ± 2.3 and 5.4 ± 2.8 for HIGH and LOW pain respectively). The result of this comparison is used as a mask for the following group analysis.
When we calculated the fMRI signal contrast between post- and pre- treatment pain application on nocebo sites from the same difference on control sites (e.g., nocebo (post-pre) − control (post-pre)), significant brain activation changes were observed in bilateral dorsal ACC, insula, superior temporal gyrus; left frontal and parietal operculum, medial frontal gyrus, orbital prefrontal cortex, superior parietal lobule and hippocampus; right claustrum / putamen, lateral prefrontal gyrus and middle temporal gyrus (Table 1). Figure 2B presents representative brain regions and corresponding cluster averaged beta values for IP noxious stimuli applied before and after treatment on nocebo and control sites, as well as for LOW and HIGH pain stimuli applied during the pre-treatment RP sequence. As indicated in the figure, pre-treatment RP HIGH stimuli produced significantly greater fMRI signal increases than RP LOW stimuli at these cluster locations. In response to identical temperature post-treatment IP pain stimuli, all fMRI signals increased on nocebo sites and decreased on control sites in these brain regions.
When we calculated the contrast by subtracting fMRI signal differences between post-and pre- treatment control sites from the same difference on nocebo sites (e.g., control (post-pre) − nocebo (post-pre)), no brain regions passed the significance threshold.
Table 2 presents the results of a simple regression analysis between each subject’s pain rating differences and corresponding fMRI signal difference between post- and pre-treatment differences on nocebo and control sites (e.g., nocebo (post-pre) − control (post-pre)). Significant positive correlations were observed in bilateral insula / frontal operculum and left M1. Significant negative correlations were observed in bilateral dorsal lateral prefrontal cortex (DLPFC) and left orbital prefrontal cortex (OPFC).
In this study, we used ROIs created from the left frontal operculum activation and left hippocampus activation as seed regions for the analysis of functional connectivity in the spontaneous, resting state fMRI scan collected prior to start of the nocebo experimental procedures. We chose left frontal operculum because it positively correlated with subjective pain rating changes and showed significantly greater fMRI signal changes on nocebo sites as compared with control sites in fMRI group analysis. (Please note the exact locations in these two analyses overlapped.) We selected the left hippocampus as a seed region because it: 1) showed significant differences in group analysis; 2) played a reported role in the correlation between anxiety and pain intensity ratings, as indicated by a previous study (Ploghaus et al., 2001) ; 3) has not been reported in the substantial placebo analgesia neuroimaging literature base (Petrovic et al., 2002; Lieberman et al., 2004; Wagner et al., 2005; Zubieta et al., 2005; Bingel et al., 2006; Kong et al., 2006b; Craggs et al., 2007; Price et al., 2007; Scott et al., 2007; Wager et al., 2007; Scott et al., 2008) and may thus be uniquely involved in nocebo hyperalgesia.
Of the 13 subjects constituting the fMRI group analysis, resting status data was only available for 12—one subject’s data was unavailable due to a technical scanning problem. To address this discrepancy in subject number, the same behavioral and fMRI group analyses were rerun using the 12 subject cohort. Identical results were found showing significant pre- and post- treatment subjective pain rating differences between nocebo and control sites (Wilcoxon Signed Ranks test, p < 0.037). fMRI analysis showed both left frontal operculum and hippocampus surviving reanalysis using the same threshold as before, while the cluster size for both regions decreased. Based on these calculations, we chose to use whole clusters as seed regions for spontaneous fMRI analysis of the 12 subjects. The details for these seed regions (coordinate, z value and cluster size) are as follows: left insula / operculum (−50 2 8; z =3.10; 38 voxels) and left hippocampus (−34 −18 −14; z = 3.84; 21 voxels).
The results of spontaneous fMRI analysis using left insula / operculum and hippocampus as seed regions are shown in Figure 3.and Table 3. For left operculum / insula, regions included bilateral operculum / insula / pre- & post-central gyrus / inferior parietal lobule/ inferior frontal gyrus / superior temporal gyrus / hippocampus / putamen, bilateral ACC / medial frontal gyrus, DLPFC, brain stem, left S1 / M1, right midcingulate and posterior cingulate cortex. For left hippocampus, brain regions included bilateral hippocampus, superior, middle & inferior temporal gyrus, insula / frontal & parietal operculum, post- & pre-central gyrus, putamen, ACC, and right orbital prefrontal cortex.
In this study, we found that after invoking nocebo effects through the creation of negative expectancy to a sham treatment, subjective pain ratings (post- minus pre-) increased significantly more on nocebo sites of the arm as compared to control sites. fMRI analysis showed brain regions involved in hyperalgesic nocebo effect during pain administration to include bilateral dorsal ACC, insula, superior temporal gyrus; left frontal and parietal operculum, medial frontal gyrus, orbital prefrontal cortex, superior parietal lobule and hippocampus; right claustrum / putamen, lateral prefrontal gyrus and middle temporal gyrus. Further analysis of spontaneous fMRI data collected before pain application showed functional connections among left frontal operculum and hippocampus and the brain regions belonging to the pain network, including bilateral insula, operculum, ACC and left M1.
We found the brain regions preferentially activated during nocebo hyperalgesic pain administration, including bilateral ACC, insula, left orbital frontal cortex, and right lateral prefrontal cortex, to reside primarily in the medial system of the pain matrix. This result is consistent with our hypothesis that nocebo hyperalgesia is predominantly produced though the affective-cognitive pain pathway. Previous studies have characterized the behavioral response to nocebo hyperalgesia and described the important role of anxiety in this process (Benedetti et al., 2006). We believe our work is the first to elucidate the brain network underlying nocebo hyperalgesia.
Previous brain imaging studies have indicated that expectation can significantly modulate subsequent noxious stimuli perception (Sawamoto et al., 2000; Koyama et al., 2005; Keltner et al., 2006). For instance, Sawamoto and colleagues (Sawamoto et al., 2000) found that uncertain expectation regarding impending painful stimuli could enhance brain responses to non-painful stimuli, increasing the intensity and range of activation in the ACC and parietal operculum and posterior insula respectively. In a later study, Keltner and colleagues (Keltner et al., 2006) found that expectancy to higher levels of pain could significantly increase reported pain intensity ratings and enhance activation of afferent pain circuits in the ipsilateral ACC, caudate, cerebellum and contralateral nucleus cuneiformis. They further hypothesized that facilitation of the descending pain modulation pathway may be involved in this process. Although not completely the same, these studies are partly consistent with our finding that during nocebo hyperalgesia, brain activity in bilateral ACC, insula and operculum on nocebo sites increases significantly more than on control sites.
Recently, spontaneous brain activity has been used to investigate functional connectivity among different brain regions (Raichle and Mintun, 2006; Buckner and Vincent, 2007; Fox and Raichle, 2007). In an early study, Biswal and colleagues (Biswal et al., 1995) found that while subjects were at rest, spontaneous fMRI BOLD signal fluctuations observed in left sensory motor areas showed a high degree of temporal correlation with right sensory motor-related brain areas and medial motor areas. This study has been replicated and findings extended to many other brain systems (Fox and Raichle, 2007). Consistent with these findings, we found highly symmetric left and right side connectivity correlations for both seed regions. Furthermore, in addition to the brain regions surrounding each seed, many regions observed in this analysis belonged to the pain network, suggesting that connectivity persists in the absence of pain stimuli as well.
In this experiment, we observed left frontal operculum activation in both group analysis and regression analysis, indicating this region’s role in nocebo hyperalgesia. Previous studies implicate operculum / insula as the most reliable region in brain imaging studies on pain (Peyron et al., 2000), and report its direct association with S1, SII, prefrontal areas, superior temporal gyrus, amygdaloid, and perirhinal cortex which is an important source of hippocampal and ACC afferents (Augustine, 1996; Cipolloni and Pandya, 1999). The above connections link brain regions to the somatosensory, limic / paralimbic, and working memory systems, providing an anatomic basis for the multiple functions and extensive functional connectivity observed in spontaneous fMRI data.
Our study also showed nocebo-induced fMRI signal changes in left hippocampus, a region known to play an important role in encoding relations between various learning context cues (Olsson and Phelps, 2007) and mediating aversive drive and the affective characteristics of pain (Melzack and Casey, 1968). The left hippocampus has been previously reported in fMRI studies on pain and anxiety. In one such study, Ploghaus and colleagues (Ploghaus et al., 2001) investigated brain response to identical pain at varying anxiety levels, observing a relationship between greater anxiety and higher pain intensity ratings. They also found left hippocampus to be uniquely involved in this process and reported that, during anxiety-induced emotional pain modulation, hippocampal responses can predict activity in closely connected, affective (perigenual cingulate), and intensity coding (mid-insula) areas. This study (Ploghaus et al., 2001) indicates that, during states of heightened anxiety, the hippocampus can amplify aversive events so as to prime behavioral responses that are adaptive for dealing with the worst possible outcome. We speculate similar mechanism may also underlie nocebo hyperalgesia.
To further evaluate the hippocampus’s role in hyperalgesic nocebo, we performed a correlation analysis between subjects’ cluster beta values for left hippocampus and other brain regions as shown in Table 1. We found that left orbital prefrontal gyrus (p = 0.003, r = 0.76) and right dACC (p = 0.024, r = 0.62) were significantly correlated with left hippocampus brain activity. The lateral orbital prefrontal cortex and dACC are known to play key roles in cognitively modulating the emotional components of pain (Petrovic and Ingvar, 2002) and processing affective aspects of pain (Price, 2000) respectively.
Interestingly, although the brain imaging literature for placebo analgesia is quite robust (Petrovic et al., 2002; Lieberman et al., 2004; Wagner et al., 2005; Zubieta et al., 2005; Bingel et al., 2006; Kong et al., 2006b; Craggs et al., 2007; Price et al., 2007; Scott et al., 2007; Wager et al., 2007; Scott et al., 2008), no placebo analgesia study to date has ever reported involvement of the hippocampus. Functional connectivity analyses using the hippocampus as a seed region demonstrate its widespread connection to pain matrix brain regions, including bilateral insula / operculum, ACC, superior parietal lobule, left M1, and pre-motor areas. This result provides further support for a possible unique role of the hippocampus in mediating nocebo hyperalgesia compared with placebo analgesia.
In this experiment, we found a significant fMRI signal increase to pain in bilateral ACC, a key region involved in processing the affective components of pain (Rainville et al., 1997; Price, 2000; Rainville, 2002). Interestingly, when we preformed the functional connectivity analysis using left frontal operculum and hippocampus as seed regions, bilateral ACC activity was observed in both cases. Signal patterns were adjacent or overlapped with fMRI group analysis results (Figure 2 and and3).3). These results further support the role of affective processing in nocebo hyperalgesia.
Our study found activity in left orbital prefrontal cortex (PFC) and right DLPFC for fMRI group analysis. A negative correlation between fMRI signal change and subjective ratings was also observed in bilateral DLPFC and left OPFC. We speculate that activity changes in PFC and parietal lobule may imply multiple functions, including memory retrieval of previous experience, expectation generation, modulation of pain perception and pain ratings, as well as attention and emotion modulation (Benedetti et al., 2006; Kong et al., 2007).
In a recent Positron Emission Tomography study, Scott and colleagues (Scott et al., 2008) asked subjects to undergo a 20 minute pain challenge and found placebo-enhanced opioid neurotransmission in the anterior cingulate, orbitofrontal and insular cortex, nucleus accumbens, amygdala and periaqueductal gray, as well as dopamine activation (DA) in the ventral basal ganglia, including the nucleus accumbens. In the same study, five subjects responding negatively to the pain challenge showed opposite changes in brain activity, a deactivation of DA and decreased opioid release in brain regions mentioned above. According to this finding, nocebo-related brain regions, including anterior cingulate, orbitofrontal and insular cortex, nucleus accumbens, and amygdale are all important limbic regions related to the interaction of emotion and pain (Vogt, 2005). Although the nature of this study may differ from our own (please note that in Scott et al.’s study, which we believed was originally designed to test placebo effect, subjects were told they would receive either an active analgesic drug or placebo, and thus the relation of these findings to nocebo effects may be limited), we similarly found nocebo hyperalgesia to exert its effects through the affective component of pain network. In addition, our work also further indicates a more extensive network of brain regions involved in nocebo hyperalgesia, including the lateral prefrontal cortex, parietal lobule and left hippocampus.
In conclusion, we found evidence that the nocebo hyperalgesic effect may be produced through the medial system of the central pain matrix responsible for affective / emotional and cognitive aspects of pain perception. Analysis of spontaneous fMRI data, collected in the absence of and preceding any pain stimuli, showed a functional connection among the brain regions observed in the subsequent nocebo scans. The left hippocampus may play an important role in nocebo hyperalgesia.
Funding and support for this study came from: KO1AT003883 to Jian Kong, NIH (NCCAM) PO1-AT002048 to Bruce Rosen, K24AT004095 to Ted Kaptchuk, R21AT00949 to Randy Gollub, M01-RR-01066 for Mallinckrodt General Clinical Research Center Biomedical Imaging Core, P41RR14075 for Center for Functional Neuroimaging Technologies from NCRR and the MIND Research Network, DE- FG03- 99ER62764 to Bruce Rosen.