PAIN�

153 (2012) 1301–1310

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

Enhanced affect/cognition-related brain responses during visceral placeboanalgesia in irritable bowel syndrome patients

Hsing-Feng Lee b,d, Jen-Chuen Hsieh a,c,d, Ching-Liang Lu a,b,d,⇑, Tzu-Chen Yeh a,c,d, Cheng-Hao Tu c,Chou-Ming Cheng c, David M. Niddam a,c,d, Han-Chieh Lin b,d, Fa-Yauh Lee b,d, Full-Young Chang b,d

a Institute of Brain Science, National Yang-Ming University, Taipei, Taiwanb Division of Gastroenterology, Taipei Veterans General Hospital, Taipei, Taiwanc Integrated brain Research laboratory, Department of Medical Research and Education, Taipei Veterans General Hospital, Taipei, Taiwand School of Medicine, National Yang-Ming University, Taipei, Taiwan

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e i n f o a b s t r a c t

Article history:Received 17 January 2011Received in revised form 23 February 2012Accepted 20 March 2012

Keywords:Placebo analgesiaVisceral painIrritable bowel syndromeFunctional brain imaging

0304-3959/$36.00 � 2012 International Associationhttp://dx.doi.org/10.1016/j.pain.2012.03.018

⇑ Corresponding author at: Institute of Brain SUniversity, Division of Gastroenterology, Taipei VeterTaiwan. Tel.: +886 2 2875 2111x3385; fax: +886 2 28

E-mail addresses: cllu@ym.edu.tw, cllu@vghtpe.go

Placebo analgesia is a psychosocial context effect that is rarely studied in visceral pain. Patients withirritable bowel syndrome (IBS) exhibit visceral hyperalgesia and heightened affective/cognitive brainregion activation during visceral stimuli. Psychological factors alter the pain and brain activation pattern,and these changes are more pronounced in IBS patients. Expectation constitutes the major neuropsycho-logical mechanism in the placebo effect. This study confirmed the heightened affective/cognitive brainresponses in IBS patients during visceral placebo analgesia using a placebo model with expectation,which was enhanced by suggestion and conditioning. Seventeen IBS patients and 17 age-/sex-matchedcontrols were enrolled. Psychophysical inventories (Hospital Anxiety and Depression Scale [HADS], visualanalogue scale, and short-form McGill questionnaire) were completed. Brain activity during placebointervention and anticipation was assessed in response to rectal distension using 3T-functional magneticresonance imaging. Suggestion-/conditioning-enhanced placebo was used to convince controls/patientsof the efficacy of a newly developed intravenous drug (saline, in actuality) for the relief of rectal disten-sion-induced visceral pain. A comparable visceral placebo analgesia was observed in IBS patients andcontrol subjects. IBS patients demonstrated a higher HADS-anxiety score, which was predictive of a weakplacebo effect. Suggestion-/conditioning-enhanced placebo evoked more activity in affective/cognitivebrain regions (insula, midcingulate cortex, and ventrolateral prefrontal cortex [VLPFC]) in IBS patientsthan in healthy controls. VLPFC was also more active during anticipation in IBS patients. In conclusion,IBS patients and control subjects achieved comparable placebo analgesia during experimentally inducedrectal pain. The visceral placebo analgesia produced heightened activity in affective/cognitive brainregions in IBS patients.

� 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

Placebo analgesia is the most well-studied placebo effect.Placebo analgesia is a psychobiological phenomenon that is attrib-uted to the ‘‘suggestion’’ of clinical improvement and Pavlovianconditioning. Placebo analgesia is mediated through opioidergicand dopaminergic pathways [42]. In the opioidergic pathway, paininhibition signals that originate in the cerebral cortex project to thehypothalamus, periaqueductal gray, and rostroventromedialmedulla before terminating in the spinal cord [41,45]. The

for the Study of Pain. Published by

cience, National Yang-Mingans General Hospital, Taipei,73 9318.

v.tw (C.-L. Lu).

dopaminergic pathway, which is responsible for reward expecta-tion, originates in the ventral tegmental area and projects to thenucleus accumbens (NAcc) [38]. These pathways were obtainedfrom ‘‘somatic’’ pain models in healthy volunteers.

Visceral and somatic pain are fundamentally different in severalaspects. Visceral pain is more poorly localized and is moreunpleasant than perceived intensity-matched somatic pain [7].Brain-imaging studies demonstrate that cortical specialization inthe sensory-discriminative and affective/cognitive areas of the cor-tex may account for the observed perceptual differences betweensomatic and visceral pain [4]. Irritable bowel syndrome (IBS) is a dis-order of chronic ‘‘visceral’’ pain in the absence of detectable disease[1]. Clinical trials of chronic visceral pain treatment in IBS patientshave revealed a placebo efficacy of approximately 50% [11,35].The mechanisms underlying the placebo response remain unclear.

Elsevier B.V. All rights reserved.

1302 H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310

Functional brain imaging has provided a window for the explo-ration of the mechanism of placebo analgesia in IBS [6,41]. Positronemission tomography has revealed an enhanced activity in theright ventrolateral prefrontal cortex (VLPFC) and an attenuatedactivity in the dorsal anterior cingulate cortex (dACC) during vis-ceral placebo analgesia in IBS patients [22]. Functional magneticresonance imaging (fMRI) revealed that the application of an inertrectal jelly, which is suggested as a pain killer, significantly de-creases brain responses in the visceral pain matrix (thalamus, insu-la, and the dACC) in IBS patients [32]. Both of these studies wereperformed only in IBS patients, and whether IBS patients andhealthy subjects share a common neural network during visceralplacebo analgesia is not known.

IBS patients display a greater activation of affect/cognition brainregions (insula, cingulate gyrus, and prefrontal cortex [PFC]) thancontrols during rectal balloon distension [34]. Psychological factors(ie, anxiety/depression or stress) modulate visceral pain more inIBS patients than in control subjects [15,16]. Expectation is the ma-jor psychological mechanism related to placebo effects [18,20,21].Expectation-related placebo analgesia in response to acupunctureis mediated through the affective/cognitive brain circuit, which in-cludes the ACC, PFC, insula, and amygdala [18,20]. These data maysuggest that expectation would induce a greater number of affec-tive/cognitive brain regions in IBS patients during placebo analge-sia. We established a placebo model with strong expectation usingthe manipulation of both ‘‘verbal suggestions’’ and ‘‘conditioning’’to produce a potent visceral placebo analgesia [26]. This studywas aimed to confirm that suggestion-/conditioning-enhanced pla-cebo involved affective/cognitive brain responses to rectal pain,which contributed to visceral placebo analgesia in IBS patients.

2. Materials and methods

2.1. Study population

Seventeen Rome III-defined IBS patients (11 female; aged35.9 ± 10.8 years) whose diagnosis had been established for morethan 1 year were recruited from our outpatient gastroenterologyclinic [23]. An additional 17 age- and sex-matched healthy volun-teers (11 female; 37.4 ± 10.2 years) were recruited through publicadvertisements. General exclusion criteria included age >55 yearsor <18 years, a body mass index >30, any concurrent medical ill-ness, including cardiovascular, endocrine, neurological conditionsor medications, with known effects on visceral sensation and brainfunction (eg, neuroleptics, antipsychotics). All subjects were evalu-ated using a digital examination for perianal tissue damage (eg,painful hemorrhoids, or anal fistula/fissure) that might interferewith balloon placement. Women in the study were tested duringthe mid-proliferative phase of their menstrual cycle to minimizevariation in pain sensitivity across the menstrual cycle. Beforethe experiment, written informed consent was obtained fromevery participant. None of the healthy volunteers were takingany medications, and the IBS patients were asked to discontinueany medications that affect gastrointestinal motor function or painperception at least 48 hours prior to the study. No IBS patients hadtaken antidepressants or anticonvulsant drugs within 1 year beforeenrollment in the current study. No participants had diabetes mel-litus, neuropsychiatric illness, or any contraindications to MRI. Allexperiments were conducted in accordance with the Declaration ofHelsinki, and the experimental protocol was approved by the Insti-tutional Review Board of Taipei Veterans General Hospital.

2.2. Experimental protocol

The protocol was similar to our previous study investigating theplacebo effect in esophageal pain [26]. Briefly, participants were

asked to fast overnight. They all completed a validated version ofthe Hospital Anxiety and Depression Scale (HADS) questionnaire[24]. The HADS consists of 14 items and was developed to assesssymptoms of anxiety and depression in the medical population[44]. The items are scored on a 4-point scale ranging from 0 (noproblem) to 3 (severe problem). In addition to the total score(HADS-total), the HADS was divided into two subscales: anxiety(HADS-A) and depression (HADS-D), each containing 7 items. A24G intravenous catheter with a normal saline lock was insertedin the left forearm of every participant. A balloon (polyethylenebag, 10 cm in length, 6 cm in diameter, with a maximum volumeof approximately 550 mL) affixed to a double-lumen polyvinyl rec-tal catheter (MUI Scientific, Mississauga, ON, Canada) was used forrectal distension. Prior to and at the end of every experiment, theballoon was inflated and deflated to ensure no leakage. The balloonwas inserted into the rectum after lubrication, with the bag margin5 cm beyond the anal verge.

In the MRI room, the balloon tube was connected via a 5-meterextension conduit (polyurethane, internal diameter 5 mm) to acomputer-controlled barostat (Distender series II; G&J, ElectronicsInc, Toronto, ON, Canada), which was placed outside of the scan-ning room.

While lying inside of the MRI scanner, but prior to commence-ment of image acquisition, participants completed threshold deter-mination (calibration) and suggestion/conditioning manipulationexperiments. The thresholds of the different sensation levels weredetermined based on a stepwise (4 mm Hg) increase in balloonpressure, as previously described [25,26]. The perception thresholdwas defined as the pressure at which participants initially felt theballoon distend. Mild and moderate pain thresholds were subse-quently determined when the participants felt 30%-40% and 60%-70% of the maximum imaginable intensity of pain. The participantswere instructed to use a response pad to measure the pain inten-sity on a homemade, MRI-compatible 100-grade digital visual ana-logue scale meter (VAS; 0 = no pain and 100 = the maximumimaginable intensity of pain), which was projected onto a screeninside the MRI room. Before the true experiment started, partici-pants performed a training session to familiarize them with theexperiment and minimize their anxiety, which could confoundthe results. As previously described [25,26], the stimulation pres-sure was further finely adjusted according to the individual’s psy-chophysical responses using continuous VAS recording. Thesubjects were given one block of mild pain and one block of mod-erate pain. The time lag between the first perception of distensionand the time at which balloon inflation began was determined forincorporation into the general linear model (described below).

To enhance the placebo effect, ‘‘verbal suggestion’’ and ‘‘condi-tioning’’ of rectal balloon-induced abdominal pain were performedas described in the previous placebo studies [26,41]. First, in pla-cebo session, we informed the volunteers and patients that theywould receive an intravenous injection of a newly developed anal-gesic that had a rapid onset and a short duration to alleviateabdominal pain induced by rectal distension of moderate intensity.Next, 5 mL of normal saline was administered intravenously, fol-lowed by rectal distension at the pressure of perception threshold(first sensation). Control session started after a resting periodaround 10 minutes (to match the time required to characterizepain in the real fMRI experiments), the participants were informedthat another intravenous injection of 5 mL normal saline withoutanalgesic effects would be given for comparison purpose. It wasfollowed by administration of rectal balloon distension eliciting amoderate level of pain intensity, a pressure which elicited pain thatwas described as 60%-70% of the maximum imaginable pain inten-sity (moderate pain). During this period of suggestion/condition-ing, the intensity of pain was obtained by direct questioningfrom each participant immediately after placebo or control

H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310 1303

sessions using a 0-100 VAS scale (0 = no pain and 100 = the maxi-mum imaginable intensity of pain), similar to that used in the cal-ibration period to determine threshold for first sensation and mild/moderate pain levels. Based on such psychophysical data (see theResults section and Table 1), we were convinced that the studiedsubjects believed in the efficacy of the ‘‘newly developed drug’’after the ‘‘suggestion/conditioning’’ manipulation.

The participants completed 4 sessions of tests in a randomizedmanner: 2 placebo and 2 control sessions. During the placebo ses-sions, they were informed that 5 mL of a newly developed analge-sic (saline) would be administered intravenously for painalleviation before rectal distension to evoke a moderate level ofpain. After this standard statement and before rectal distension,there was an anticipation period of 60 seconds. The fMRI imageswere acquired in synchronization with the rectal distension fol-lowing the injection of the newly developed ‘‘drug.’’ Each placebosession consisted of 2 30-second blocks of rectal distension alter-nating with 30 seconds of the baseline condition (no balloon dis-tension). Two placebo sessions were conducted consecutively(Fig. 1). The control sessions were organized identically to the pla-cebo sessions. However, participants were informed that 5 mL ofnormal saline would be administered and followed by distensioninducing a similar intensity of pain. The protocol was not explainedin detail to the participants before the study; thus, they did notknow the exact duration of the anticipation or rectal distensionphases. The order of placebo and control sessions was balancedacross participants. After completing 2 similar sessions, partici-pants were asked to draw the spatial extent of pain experiencedon a structured, A4-sized paper, to rate their perceived painaccording to the VAS, and to fill out a validated Taiwanese versionof the short-form McGill pain questionnaire (SF-MPQ) [8]. The SF-MPQ consists of a series of questions regarding different aspects ofreported pain. The main component of the SF-MPQ consists of 15pain descriptors (11 sensory; 4 affect): 1-11 (throbbing, shooting,stabbing, sharp, cramping, gnawing, hot-burning, aching, heavy,tender, and splitting) represent the sensory dimension of painexperience, and 12-15 (tiring-exhausting, sickening, fearful, pun-ishing-cruel) represent the affective dimension. Each descriptorwas ranked on an intensity scale with the following labels:0 = none, 1 = mild, 2 = moderate, and 3 = severe. Two major indiceswere obtained for the sensory as well as for the affective compo-nents of pain: 1) the number of words chosen (NWC), and 2) thepain rating index (PRI), based on a summation of the numericalvalues assigned to the chosen words. The total score in either

Table 1Comparison of the demographics, HADS, barostat study data and suggestion/conditioning manipulation data between healthy volunteers and IBS patients.

Healthy volunteers(n = 17)

IBS patients(n = 17)

Pvalue

Age (y) 37.4 ± 10.2 36.0 ± 10.8 0.60Male/female 6/11 6/11HADS-total 10.3 ± 5.5 15.1 ± 6.5 0.04HADS-anxiety 5.8 ± 3.2 8.8 ± 3.6 0.01HADS-

depression4.5 ± 2.9 6.3 ± 3.7 0.15

Barostat study (mm Hg)First sensation 21.9 ± 5.9 17.5 ± 4.8 0.04Mild pain 35.3 ± 7.6 29.8 ± 6.8 0.01Moderate pain 48.8 ± 7.6 40.7 ± 7.9 0.008

Suggestion/conditioningPlacebo 16.0 ± 9.2* 19.4 ± 9.9* 0.22Control 64.1 ± 5.4 60.7 ± 9.7 0.36

HADS, Hospital Anxiety and Depression Scale.Presented values are mean ± SD.* P < 0.001, Placebo vs Control.

NWC or PRI is the summation of sensory and affective components.The participants were asked not to move their heads and to closetheir eyes during the fMRI scanning. They were instructed to con-centrate on the experience of rectal distension-induced pain. Theextent of pain was assessed by digitizing the drawings (image res-olution: 2550 � 3509). Because the recording paper had a size ofA4, the pixel size was calculated to be 6.97 � 10�5 cm2. The areaof pain extent was then calculated by multiplying the pixel sizeby the pixel count.

2.3. MRI scanning

Images were acquired using a 3.0 T Bruker MedSpec S300 sys-tem (Bruker, Karlsruhe, Germany) with a quadrature head coil.The subjects’ heads were immobilized in the scanner with a vac-uum pillow. Functional data were acquired with a T2⁄-weightedgradient-echo echo planar image using blood oxygen level-depen-dent (BOLD) contrast (repetition time [TR]/echo time [TE]/h = 2000 ms/50 ms/90�, slice thickness = 5 mm, interslice inter-val = 1 mm, field of view [FOV] = 230 � 230 mm2, 64 � 64 � 20matrix, whole brain coverage). For each session, the first 5 images(dummy images) were discarded from the analysis to eliminatepossible nonequilibrium effects of magnetization. An anatomicalimage was acquired using a high-resolution, T1-weighted, 3D gra-dient-echo pulse sequence (modified driven equilibrium Fouriertransform; TR/TE/TI [inversion time] = 88.1 ms/4.12 ms/650 ms,256 � 256 � 128 matrix, FOV = 230 � 230 � 192 mm).

2.4. fMRI analysis

Data were analyzed by statistical parametric mapping (SPM5software from the Wellcome Department of Cognitive Neurology,London) using Matlab 7.4 (Mathworks, Sherbon, MA, USA). Scanswere realigned, coregistered, normalized, time corrected, and spa-tially smoothed with an 8-mm full-width-at-half-maximumGaussian kernel using standard SPM methods. The predeterminedlatencies of perception onset (1.8 ± 0.8 seconds for healthy volun-teers and 1.5 ± 0.9 seconds for IBS patients) relative to the trigger-ing stimuli were incorporated as the regional cerebral blood flowresponse delay into the synthetic hemodynamic response functionmodel of SPM5. Brain activation during either control or placebosessions was determined by the contrast between balloon disten-sion and baseline. Brain activity associated with the placebo effectwas then determined based on the contrast between control andplacebo sessions following distension. Brain activity due to antici-pation of the placebo analgesia was represented by the contrast be-tween placebo and control sessions prior to rectal distension. Two-sample t-tests, as implemented in SPM5 software, were used to ad-dress group differences between IBS patients and healthy volun-teers during anticipation and rectal distension. Brain activity wasexamined by voxel-specific t-tests (SPM [t]) across all participants.The t-statistics were subsequently transformed to Z-statistics togenerate a statistical parametric map [SPM (Z)] for each contrast.We hypothesized that the visceral placebo analgesia would beassociated with dynamic changes in the BOLD response evokedthroughout the visceral pain matrix as well as in brain regionsresponsible for affect/cognition. Furthermore, a priori knowledgeis already available on the brain network that is engaged byvisceral pain and is associated with affect/cognition; these areasinclude the prefrontal, somatosensory, insula, and cingulatecortex as well as the thalamus, midbrain, and cerebellum[16,18,22,26,29,32,41]. Therefore, we selected the ‘‘small volumecorrection’’ method to further examine the reliability of the acti-vated regions observed for the uncorrected measurement[15,16,26]. Briefly, regionally specific differences that were ob-served at an uncorrected threshold of P < 0.005 (cluster size = 10

Fig. 1. Stimulation paradigm during the control or placebo sessions.

1304 H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310

voxels) were first examined [41]. After this uncorrected compari-son, significance was considered only if the regions passed a spher-ical small volume correction centered at the peak location(radius = 5 mm, corrected, P < 0.05). Z-maxima were localized onthe Montreal Neurological Institute (MNI) 152 template and la-beled by means of the Talairach Daemon (Research Imaging Center,University of Texas, San Antonio, TX, USA). In multiple regressionanalyses, fMRI data from IBS patients during visceral placebo anal-gesia (placebo � control) were correlated with HADS scores as thedependent variable. This was done only in the IBS group because acorrelation between HADS scores and the magnitude of placebo ef-fects has been previously identified in IBS patients but not inhealthy volunteers (see Results below).

2.5. Statistical analysis of demographic, barostatic, pain area extent,and psychophysical data

A Wilcoxon signed rank test or Mann-Whitney U test was usedfor within- or between-group comparisons, respectively. Spear-man’s correlation analysis was conducted to examine the relation-ship between HADS and differences in the extent of the pain areaexperienced, VAS, and SF-MPQ parameters after placebo analgesia.A P value <0.05 was considered significant.

3. Results

3.1. Baseline and psychophysical data of the placebo effect in healthyvolunteers and IBS patients

Table 1 shows baseline data for both groups. The IBS patientsdemonstrated a significantly lower visceral perception thresholdupon rectal distension at all 3 perception levels and a higherHADS-total and HADS-A, compared to controls. A lower VAS scorewas successfully induced after the ‘‘suggestion’’ and ‘‘conditioning’’manipulation in both healthy volunteers and IBS patients (Table 1).Table 2 shows that the spatial extent of pain, VAS score, and all SF-MPQ parameters, except the ‘‘NWC-affect’’ in controls, were signif-icantly reduced after placebo treatment in both healthy volunteersand IBS patients. No difference in the placebo effect was noted be-tween the 2 groups. Fig. 2A and B show the extent of the pain areaexperienced following control or placebo treatment in healthy vol-unteers and IBS patients. These results suggested that both healthyvolunteers and IBS patients could achieve a comparable placeboanalgesic effect in response to experimentally induced rectal pain.

In IBS patients, HADS-total was negatively correlated withreduction of the ‘‘PRI-sensory’’ index (r = �0.56, P = 0.02) upon pla-cebo treatment. Further analysis revealed a significantly negativecorrelation between HADS-A and reduction of ‘‘NWC-total’’(r = �0.63, P = 0.01), ‘‘NWC-sensory’’ (r = �0.62, P = 0.01), and‘‘PRI-sensory’’ (r = �0.55, P = 0.02) indices following placebo anal-gesia. These results suggested that a higher HADS might partiallypredict a weak placebo effect in IBS patients. In healthy volunteers,

no correlations were noted between HADS and any SF-MPQparameters.

3.2. Brain responses after rectal balloon stimulation in control andplacebo sessions of healthy volunteers and IBS patients

Among the healthy volunteers (Fig. 3A, B) and IBS patients(Fig. 3C, D), rectal pain during either the control or placebo sessionselicited various levels of activation in a wide range of cortical andsubcortical regions that are congruent with the traditional visceralpain matrix, including S1/S2, insula, dACC, and prefrontal cortex.

3.3. Neuronal correlates to placebo analgesia upon rectal distensionwithin healthy volunteers or IBS patients (within-group analysis)(Fig. 4A, B)

When we calculated the contrast between control and placebo(control � placebo) in healthy volunteers, we observed that BOLDresponses evoked in visceral pain-associated areas such as S1/S2,ACC, insula, and prefrontal cortices, were reduced after placebotreatment (Fig. 4A). In IBS patients, the placebo treatment induceda significant decrease in brain activity, mainly in the posterior cin-gulate cortex (PCC), bilateral thalamus, and cerebellum (Fig. 4B).

We also calculated the BOLD contrast between placebo and con-trol (placebo � control) in both groups. The NAcc and PCC showedgreater activation after placebo treatment than after control treat-ment in healthy volunteers, whereas insula, mid-cingulate cortex(MCC), bilateral VLPFC, and precuneus showed greater activationfollowing placebo treatment in IBS patients.

3.4. Group differences in the neural correlates underlying the modula-tion of placebo analgesia (placebo � control) upon rectal distension(between-groups analysis)

Direct group comparisons with 2-sample t-tests revealed brainareas in which the activation elicited by placebo analgesia differedbetween groups. The IBS vs healthy volunteer (IBS > health) con-trast identified significant differences in the bilateral VLPFC, insula,MCC, PCC, right claustrum, and precuneus (Table 3, Fig. 5). Thisfinding supported the concept that the placebo would induce morebrain regions involving affective/cognitive function, which mightlead to the visceral placebo analgesia observed in IBS patients. Onthe other hand, the healthy volunteer vs IBS (IBS < health) contrastshowed that IBS patients had less activation in the cerebellum andmidbrain than controls during placebo treatment (Table 3, Fig. 5).

3.5. Neuronal correlates during anticipation of placebo analgesia(placebo � control) prior to rectal distension between healthyvolunteers and IBS patients (between-group analysis)

As shown in Figs. 6 and 2 sample t-tests showed that the IBSgroup had higher neuronal activation in left VLPFC (Brodmann area

Table 2Comparison of the extent of the pain area experienced, VAS, and short-form McGill pain questionnaire (SF-MPQ) between control and placebo sessions in healthy volunteers andIBS patients.

Healthy volunteers (n = 17) IBS patients (n = 17)

Control Placebo Placebo effect Control Placebo Placebo effect

Pain area (cm2) 7.7 ± 5.1 4.3 ± 3.2 3.4 ± 2.7 a 8.7 ± 6.9 5.1 ± 6.3 3.6 ± 5.3a

VAS score 63.7 ± 7.3 43.0 ± 15.0 20.6 ± 14.5a 57.5 ± 13.3 40.8 ± 15.2 16.7 ± 15.8a

SF-MPQPRI-total 15.3 ± 7.9 9.4 ± 6.6 5.9 ± 6.2a 14.5 ± 7.0 8.3 ± 5.2 6.2 ± 6.0a

PRI-sensory 12.6 ± 5.9 7.7 ± 5.1 4.9 ± 4.9a 12.5 ± 5.8 7.4 ± 4.2 5.1 ± 5.0a

PRI-affective 2.7 ± 2.7 1.7 ± 2.3 1.1 ± 1.8a 2.0 ± 2.1 0.9 ± 1.4 1.1 ± 1.6a

NWC-total 8.2 ± 3.7 6.2 ± 3.2 2.1 ± 1.8a 8.4 ± 3.4 6.7 ± 3.4 1.7 ± 2.0a

NWC-sensory 6.6 ± 2.6 5.1 ± 2.4 1.5 ± 1.5a 7.0 ± 2.6 5.8 ± 2.5 1.2 ± 1.6a

NWC-affective 1.7 ± 1.4 1.1 ± 1.2 0.5 ± 1.5 1.4 ± 1.2 0.9 ± 1.2 0.5 ± 0.8a

VAS, visual analogue scale; IBS, irritable bowel syndrome; PRI, pain rating index; NWC, number of words chosen.a Control vs Placebo, P < 0.05.

Fig. 2. The extent of the pain area induced by rectal balloon distension during control and placebo sessions in (A) healthy volunteers and (B) irritable bowel syndrome (IBS)patients. The numbers on the color bar indicate the number of subjects with an overlapping extent of pain.

H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310 1305

(BA) 47, x = �44, y = 24, z = �6, z-score = 2.88) than healthy volun-teers (IBS > health) during placebo anticipation (placebo � con-trol). No significant clusters were found for the (IBS < health)contrast during anticipation.

3.6. Correlation between HADA-A and brain activity in placeboanalgesia (placebo � control) in IBS patients

Because higher HADS-A scores have been identified to be pre-dictive of a weak placebo effect in IBS patients, the correlation

between HADS-A and brain activity during placebo analgesia (pla-cebo � saline) was further analyzed using multiple regressionanalyses in the IBS group. HADS-A was negatively correlated withactivation in the left VLPFC/insula (BA 47, x = �36, y = 22, z = �14,z-score = 3.69) in IBS patients (Fig. 7).

4. Discussion

The present study demonstrated that IBS patients achieved pla-cebo analgesia comparable to that in controls during experimentally

Fig. 3. Baseline brain activation upon rectal balloon stimulation during control and placebo sessions in healthy volunteers and irritable bowel syndrome (IBS) patients.Healthy volunteers (HV) in (A) control and (B) placebo sessions; IBS patients in (C) control and (D) placebo sessions. Variable degrees of activation are observable in corticaland subcortical regions congruent with the visceral pain matrix, mainly S1/S2, insula, dorsal anterior cingulate cortex, and prefrontal cortex. Statistical images arecoregistered with the SPM-MNI template in Talairach space. The color bar indicates the z-score associated with the activation. Axial views are displayed according toneurological conventions (ie, the subject’s right hemisphere is shown on the right side of the images).

1306 H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310

induced rectal pain. However, different brain circuits were involvedin visceral placebo analgesia in IBS patients compared to controls.More affective/cognitive brain regions were engaged in the modula-tion of visceral placebo analgesia in IBS patients.

Although high placebo effect is almost always observed in ‘‘clin-ical trials’’ among IBS patients, no previous study has attempted tocompare visceral placebo analgesia elicited in healthy volunteersand IBS patients. In this study designed to explore placebo mecha-nism, we observed a similar magnitude of placebo effect betweenthe 2 groups. We speculated that the ‘‘verbal suggestion’’ and ‘‘con-ditioning’’ before the fMRI experiment would greatly enhance theexpectation of pain relief in both groups; such maneuvers mightmake the magnitude of the placebo response too large to detectdifferences between the 2 groups. In addition, we showed that aplacebo response is more easily induced in visceral pain than so-matic pain, a finding that might also contribute to this observation[26]. Actually, this similar pain relief after placebo treatmentwould help to better characterize the fundamental differences inthe brain responses between the 2 groups. Also, our fMRI datashowed that IBS patients displayed different neuro-circuits duringplacebo processing despite similar placebo effects when comparingto the control. This finding would further support a role for dysreg-ulation of the central nervous system in IBS patients.

Affective/cognitive brain regions (insula, MCC, and VLPFC)exhibited greater activation during visceral placebo analgesia inIBS patients. The insula is associated with interoception, which isthe integration of the current state of bodily processes and theaffective component of pain [3,10]. For example, the insula is in-volved in the stimulation-induced reduction of pain and the mod-ulation of pain perception by a negative emotional state [30]. Theinsula also modulates analgesic pain during placebo analgesia(for a review, see [45]). The MCC receives numerous direct inputsfrom the amygdala [40], which is implicated in fear processingand nociception [5,43]. The MCC may mediate the fear avoidanceaspect of pain processing specifically because fear and pain pro-cessing overlap in this area [12]. The VLPFC participates in the cog-nitive modulation of pain, including the visceral placebo analgesiathat was observed in our previous study [26]. These data suggest areappraisal of the significance of a stimulus or the effects of emo-tional context on cognitive evaluation processes. Our results aresupported by a previous placebo study that demonstrated a similar

brain involvement in the ACC, insula, and VLPFC during sham acu-puncture for thermal pain relief [18].

Mechanisms underlying the increased activation of the insula,MCC, and VLPFC during visceral placebo analgesia in IBS patientscompared to controls are not known. We postulated that the en-hanced expectation in the current study design is one of the factorsthat is responsible for this observation. IBS patients exhibited vis-ceral hyperalgesia with aberrant brain activation during visceralstimuli [34,37]. Psychological factors affect visceral pain and theassociated brain responses more in IBS patients than controls[15,16,27]. For example, acute stress-induced depressive symp-toms and negative emotions contribute to altered neural responsesto visceral stimulation, and IBS patients exhibit greater alterationsthan control subjects [15]. Affective/cognitive brain regions weremore activated in IBS than control during rectal distension, andthese differences were lost when affective symptoms (eg, anxi-ety/depression) were considered as confounds [16]. Placebo alsomodulates visceral pain in IBS [33]. And psychological factors, suchas expectation, but not endogenous opioid mechanisms, determinethe placebo response in IBS patients [39]. Furthermore, expectationalso determines acupuncture-evoked placebo analgesia[18,20,21,39]. The current study manipulated the placebo responseby the implementation of a verbal suggestion and a prior learningexperience (ie, conditioning) to induce strong expectations and apotent analgesic effect. This psychological manipulation shouldheighten the affective/cognitive component in the pathophysiol-ogy of IBS patients compared to controls. The IBS patients mightthen display a stronger desire for pain relief than controls underthis manipulation. Besides the potential effect of expectation onthe increased brain responses in IBS patients, the heightened brainactivation might also serve as a counterbalancing or coping mech-anism to enable IBS patients to achieve a placebo effect comparableto that in controls [17].

Greater VLPFC activation was noted in IBS patients than in thecontrols during the anticipation stage of placebo analgesia. TheVLPFC undergoes placebo-dependent activation that is correlatedwith the placebo effect and treatment expectation [28]. VLPFCactivation is observed during the anticipation of esophageal pain[26]. Therefore, VLPFC activation might be integral to the mainte-nance of the expectations of pain relief and the use of theseexpectations as a context for the diversion of attention away from

Fig. 4. Within-group differences during placebo analgesia in (A) healthy volunteers (HV) and (B) irritable bowel syndrome (IBS) patients. (A) In healthy volunteers, placebotreatment induced significantly decreased neuronal activity mainly in components of the visceral pain matrix, such as anterior cingulate cortex (ACC) (x = �8, y = 24, z = 24),right insula (x = 42, y = 20, z = �6), prefrontal cortices (x = �54, y = 18, z = 18) and somatosensory cortices (x = 54, y = �36, z = 26) [control > placebo]. Reverse calculationshowed that the nucleus accumbens (NAcc, x = 4, y = 2, z = �2) and posterior cingulate gyrus (PCC) (x = 8, y = �56, z = 20) were activated after placebo treatment[placebo > control]. (B) In IBS patients, enhanced activity in affective/cognitive brain regions such as bilateral ventrolateral prefrontal cortex (VLPFC; x = 32, y = 22, z = �8;x = �44, y = 20, z = �2), left insula (x = �46,y = �8, z = 4), mid-cingulate cortex (MCC) (x = 20, y = �14, z = 46), and precuneus (x = 28, y = �62, z = 24) was found with the[placebo > control] contrast. Reverse calculation showed enhanced activation in PCC (x = 6, y = �44, z = 6), the thalamus (x = 0, y = �20, z = 8; x = 8, y = �22, z = 2), themidbrain (x = 14, y = �22, z = �8), and the cerebellum (x = 18, y = �60, z = �14) [control > placebo].

H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310 1307

pain [14,19]. This result is consistent with our speculation that IBSpatients tended to display more desire for pain relief thancontrols.

A negative correlation between HADS-A and the placebo re-sponse was observed only in IBS patients. IBS patients often exhibitmood disorders, such as anxiety and depression [1,13]. High anxi-ety was predictive of weak visceral placebo analgesia in IBS pa-tients, but the underlying mechanisms for this finding are notclear. IBS patients with high anxiety might be more suspicious ofthe efficacy of placebo treatment, which could create less desireand expectation of pain relief. A positive correlation betweenpreoperative anxiety and postoperative pain severity has been re-ported previously [36]. In addition, higher baseline anxiety anddepression levels measured by HADS are also predictive of a poor

response to treatment in patients with functional gastrointestinaldisorder [31]. Therefore, our results that IBS patients with highanxiety levels tended to have poor placebo responses are notsurprising.

One might assume that a positive correlation, instead of thenegative correlation that was identified in our results, should beobserved between HADS-A and the placebo response because ourfMRI findings suggested an enhanced engagement of affectivebrain regions during the modulation of placebo in IBS patients.However, our data did not provide direct evidence of whether thisengagement led to a heightened or diminished placebo response.The engagement of affective components might utilize inhibitoryor other complicated mechanisms to produce a weak placeboresponse.

Table 3Difference in the neuronal correlates during analgesia (placebo � control) betweenIBS and healthy volunteers by 2-sample t-test (between-groups analysis).

Brain region H BA Peak coordinate Z-score

x y z

IBS > HealthInferior frontal gyrus L 47 �28 30 �4 2.83Inferior frontal gyrus R 44,47 38 32 4 3.34Precuneus R 7 16 �70 34 3.44Superior temporal gyrus L 22 �50 14 �2 3.29Lingual gyrus R 19 28 �70 0 2.91Insula L 13 �46 �10 4 3.23Insula R 13 34 �12 22 2.95Claustrum L �36 �22 6 3.35Claustrum R 38 �8 12 3.18Cingulate gyrus R 24 22 �16 44 2.83Parahippocampal gyrus R 30 28 �54 4 3.23Posterior cingulate R 23 12 �30 26 3.15Thalamus R 26 �22 6 2.87

IBS < HealthCerebellum R 2 �58 �2 3.15Midbrain L 0 �34 �4 3.04

IBS, irritable bowel syndrome; H, hemisphere with activation; all coordinates wereconverted from MNI to Talairach space; BA, Brodmann area.All small-volume corrected P < 0.05.

Fig. 6. Significant group differences during anticipation in irritable bowel syn-drome (IBS) patients vs healthy volunteers in the left ventrolateral prefrontal cortex(VLPFC) (x = �44, y = 24, z = �6) P < 0.05, small-volume corrected.

1308 H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310

Price et al. demonstrated greatly reduced brain activity in re-sponse to pain in IBS patients during the placebo effect [32]. How-ever, we found a decrease in neuronal activity primarily in the PCCand bilateral thalamus. In contrast, enhanced activity in the rightVLPFC is associated with attenuated activity in the dACC duringvisceral stimulation after placebo treatment [22]. We speculatedthat these studies might have measured different facets of the pla-cebo effect in IBS patients because of the great variation in studydesign. In suggestion-enhanced placebo studies, IBS patients weretold that an ‘‘effective’’ placebo would be given to relieve the vis-ceral pain through a short-term systemic (intravenous) or local(rectal) route. However, IBS patients in clinical trials were in-formed that ‘‘you might orally receive an active pain-reducingmedication or an inert placebo agent,’’ and remained blinded to

Fig. 5. Between-group difference between healthy volunteers (HV) and irritable bowincreased activation in bilateral ventrolateral prefrontal cortex (VLPFC)/insula (x = 30, y =cortex (MCC) (x = 20, y = �16, z = �4), and precuneus (x = 16, y = �70, z = 34) during placeactivation in cerebellum and midbrain for healthy volunteers compared to IBS patients.

the treatment type for the duration of treatment (3 weeks). Fur-thermore, differences in the route of administration (eg, intrave-nous, oral, or rectal) may affect the placebo response in IBSpatients. Similar differences are observed in the placebo analgesiaof somatic pain in healthy subjects. For example, placebo (cream)analgesia was associated with a decrease in brain activity inpain-sensitive brain regions, including the thalamus, insula, andACC [41]. However, the left motor cortex was the only region witha decrease in neuronal activity using sham acupuncture [18]. Weagree that placebo analgesia is a complicated phenomenon that in-volves different mechanisms under different conditions [2,9].Incoming signals from primary sensory neurons undergo extensiveassociative elaboration in the brain through cortico-subcorticaland corticocortical pathways that are modulated by brain regions,such as the PFC, cingulate gyrus, and insula [18]. Placebo analgesiais the result of a combination of different components. Differentexperimental designs in placebo studies will determine the differ-ent ratio of placebo components and lead to a varied activation ofneural circuits during visceral placebo analgesia. Therefore,

el syndrome (IBS) patients [placebo > control]. IBS patients showed significantly24, z = �8; x = �28, y = 30, z = �4), left insula (x = �46, y = �10, z = 4), mid-cingulatebo analgesia compared to healthy volunteers. Reverse calculation showed increasedPCC, posterior cingulate cortex.

Fig. 7. Correlation between Hospital Anxiety and Depression Scale-anxiety (HADS-A) and brain activity during placebo analgesia in irritable bowel syndrome (IBS)patients. HADS-A scores were negatively correlated with activation in the leftventrolateral prefrontal cortex (VLPFC)/insula (x = �36, y = 22, z = �14) P < 0.005,uncorrected.

H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310 1309

discrepancies among placebo studies may be due to the measure-ment of different components of the placebo phenomenon [18].

One might argue that our paradigm consisted of long blocks,which would be highly predictable and contaminate our placeboeffects. However, our study paradigm was unlikely predictablefor the following reasons. First, all participants were naïve to thefMRI study, and no immediate learning effect was expected to oc-cur. Second, the participants were completely unaware of the exactprotocol, and they did not know when pain would be initiated.Therefore, anticipation stage and painful stage during placeboanalgesia were independent. Nevertheless, brain mechanisms thatare involved in coping strategies may have confounded our obser-vations during the painful blocks [17].

In conclusion, IBS patients achieved visceral placebo effects torelieve experimentally induced rectal pain that were comparableto the placebo effects observed in healthy volunteers. More acti-vated affective/cognitive brain regions were noted in IBS patientsthan in controls during visceral placebo analgesia.

Conflicts of interest statement

No conflicts of interest exist among all authors regarding thismanuscript.

Acknowledgements

This study was made possible by grant support from the Na-tional Science Council of Taiwan 96-2314-B-010-024, 99-2314-B-010-015 and Taipei Veterans General Hospital v98c1-071.

References

[1] Agrawal A, Whorwell PJ. Irritable bowel syndrome: diagnosis andmanagement. BMJ 2006;332:280–3.

[2] Amanzio M, Benedetti F. Neuropharmacological dissection of placeboanalgesia: expectation-activated opioid systems versus conditioning-activated specific subsystems. J Neurosci 1999;19:484–94.

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

[4] Aziz Q, Thompson DG, Ng VW, Hamdy S, Sarkar S, Brammer MJ, Bullmore ET,Hobson A, Tracey I, Gregory L, Simmons A, Williams SC. Cortical processing ofhuman somatic and visceral sensation. J Neurosci 2000;20:2657–63.

[5] Bernard JF, Huang GF, Besson JM. Nucleus centralis of the amygdala and theglobus pallidus ventralis: electrophysiological evidence for an involvement inpain processes. J Neurophysiol 1992;68:551–69.

[6] Bingel U, Lorenz J, Schoell E, Weiller C, Buchel C. Mechanisms of placeboanalgesia: rACC recruitment of a subcortical antinociceptive network. Pain2006;120:8–15.

[7] Cervero F. Visceral versus somatic pain: similarities and differences. Dig Dis2009;27:3–10.

[8] Chang MY, Chen CH, Huang KF. A comparison of massage effects on labor painusing the McGill Pain Questionnaire. J Nurs Res 2006;14:190–7.

[9] Colloca L, Benedetti F. Placebos and painkillers: is mind as real as matter? NatRev Neurosci 2005;6:545–52.

[10] Craig AD. Interoception: the sense of the physiological condition of the body.Curr Opin Neurobiol 2003;13:500–5.

[11] Cremonini F, Delgado-Aros S, Camilleri M. Efficacy of alosetron in irritablebowel syndrome: a meta-analysis of randomized controlled trials.Neurogastroenterol Motil 2003;15:79–86.

[12] Dorn SD, Palsson OS, Thiwan SI, Kanazawa M, Clark WC, van Tilburg MA,Drossman DA, Scarlett Y, Levy RL, Ringel Y, Crowell MD, Olden KW, WhiteheadWE. Increased colonic pain sensitivity in irritable bowel syndrome is the resultof an increased tendency to report pain rather than increased neurosensorysensitivity. Gut 2007;56:1202–9.

[13] Drossman DA, Camilleri M, Mayer EA, Whitehead WE. AGA technical review onirritable bowel syndrome. Gastroenterology 2002;123:2108–31.

[14] Elsenbruch S. Abdominal pain in Irritable Bowel Syndrome: a review ofputative psychological, neural and neuro-immune mechanisms. Brain BehavImmun 2011;25:386–94.

[15] Elsenbruch S, Rosenberger C, Bingel U, Forsting M, Schedlowski M, GizewskiER. Patients with irritable bowel syndrome have altered emotional modulationof neural responses to visceral stimuli. Gastroenterology 2010;139:1310–9.

[16] Elsenbruch S, Rosenberger C, Enck P, Forsting M, Schedlowski M, Gizewski ER.Affective disturbances modulate the neural processing of visceral pain stimuliin irritable bowel syndrome: an fMRI study. Gut 2010;59:489–95.

[17] Hsieh JC, Stone-Elander S, Ingvar M. Anticipatory coping of pain expressed inthe human anterior cingulate cortex: a positron emission tomography study.Neurosci Lett 1999;262:61–4.

[18] Kong J, Gollub RL, Rosman IS, Webb JM, Vangel MG, Kirsch I, Kaptchuk TJ. Brainactivity associated with expectancy-enhanced placebo analgesia as measuredby functional magnetic resonance imaging. J Neurosci 2006;26:381–8.

[19] Kong J, Kaptchuk TJ, Polich G, Kirsch I, Gollub RL. Placebo analgesia: findingsfrom brain imaging studies and emerging hypotheses. Rev Neurosci2007;18:173–90.

[20] Kong J, Kaptchuk TJ, Polich G, Kirsch I, Vangel M, Zyloney C, Rosen B, Gollub R.Expectancy and treatment interactions: a dissociation between acupunctureanalgesia and expectancy evoked placebo analgesia. Neuroimage2009;45:940–9.

[21] Kong J, Kaptchuk TJ, Polich G, Kirsch I, Vangel M, Zyloney C, Rosen B, Gollub RL.An fMRI study on the interaction and dissociation between expectation of painrelief and acupuncture treatment. Neuroimage 2009;47:1066–76.

[22] Lieberman MD, Jarcho JM, Berman S, Naliboff BD, Suyenobu BY, Mandelkern M,Mayer EA. The neural correlates of placebo effects: a disruption account.Neuroimage 2004;22:447–55.

[23] Longstreth GF, Thompson WG, Chey WD, Houghton LA, Mearin F, Spiller RC.Functional bowel disorders. Gastroenterology 2006;130:1480–91.

[24] Lu CL, Liu CC, Fuh JL, Liu PY, Wu CW, Chang FY, Lee SD. Irritable bowelsyndrome and negative appendectomy: a prospective multivariableinvestigation. Gut 2007;56:655–60.

[25] Lu CL, Wu YT, Yeh TC, Chen LF, Chang FY, Lee SD, Ho LT, Hsieh JC. Neuronalcorrelates of gastric pain induced by fundus distension: a 3T-fMRI study.Neurogastroenterol Motil 2004;16:575–87.

[26] Lu HC, Hsieh JC, Lu CL, Niddam DM, Wu YT, Yeh TC, Cheng CM, Chang FY, LeeSD. Neuronal correlates in the modulation of placebo analgesia inexperimentally induced esophageal pain: a 3T-fMRI study. Pain2010;148:75–83.

[27] Mertz H, Morgan V, Tanner G, Pickens D, Price R, Shyr Y, Kessler R. Regionalcerebral activation in irritable bowel syndrome and control subjects withpainful and nonpainful rectal distention. Gastroenterology 2000;118:842–8.

[28] Petrovic P, Dietrich T, Fransson P, Andersson J, Carlsson K, Ingvar M. Placebo inemotional processing–induced expectations of anxiety relief activate ageneralized modulatory network. Neuron 2005;46:957–69.

[29] Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia—imaging a shared neuronal network. Science 2002;295:1737–40.

[30] Phillips ML, Gregory LJ, Cullen S, Coen S, Ng V, Andrew C, Giampietro V,Bullmore E, Zelaya F, Amaro E, Thompson DG, Hobson AR, Williams SC,Brammer M, Aziz Q. The effect of negative emotional context on neural andbehavioural responses to oesophageal stimulation. Brain 2003;126:669–84.

[31] Porcelli P, Bagby RM, Taylor GJ, De Carne M, Leandro G, Todarello O.Alexithymia as predictor of treatment outcome in patients with functionalgastrointestinal disorders. Psychosom Med 2003;65:911–8.

[32] Price DD, Craggs J, Verne GN, Perlstein WM, Robinson ME. Placebo analgesia isaccompanied by large reductions in pain-related brain activity in irritablebowel syndrome patients. Pain 2007;127:63–72.

[33] Price DD, Craggs JG, Zhou Q, Verne GN, Perlstein WM, Robinson ME.Widespread hyperalgesia in irritable bowel syndrome is dynamicallymaintained by tonic visceral impulse input and placebo/nocebo factors:evidence from human psychophysics, animal models, and neuroimaging.Neuroimage 2009;47:995–1001.

[34] Rapps N, van Oudenhove L, Enck P, Aziz Q. Brain imaging of visceral functionsin healthy volunteers and IBS patients. J Psychosom Res 2008;64:599–604.

[35] Spanier JA, Howden CW, Jones MP. A systematic review of alternativetherapies in the irritable bowel syndrome. Arch Intern Med 2003;163:265–74.

1310 H.-F. Lee et al. / PAIN�

153 (2012) 1301–1310

[36] Thomas T, Robinson C, Champion D, McKell M, Pell M. Prediction andassessment of the severity of post-operative pain and of satisfaction withmanagement. Pain 1998;75:177–85.

[37] Tillisch K, Mayer EA, Labus JS. Quantitative meta-analysis identifies brainregions activated during rectal distension in irritable bowel syndrome.Gastroenterology 2011;140:91–100.

[38] Tracey I. Getting the pain you expect: mechanisms of placebo, nocebo andreappraisal effects in humans. Nat Med 2010;16:1277–83.

[39] Vase L, Robinson ME, Verne GN, Price DD. Increased placebo analgesia overtime in irritable bowel syndrome (IBS) patients is associated with desire andexpectation but not endogenous opioid mechanisms. Pain 2005;115:338–47.

[40] Vogt BA. Pain and emotion interactions in subregions of the cingulate gyrus.Nat Rev Neurosci 2005;6:533–44.

[41] Wager TD, Rilling JK, Smith EE, Sokolik A, Casey KL, Davidson RJ, Kosslyn SM,Rose RM, Cohen JD. Placebo-induced changes in FMRI in the anticipation andexperience of pain. Science 2004;303:1162–7.

[42] Watson A, El-Deredy W, Iannetti GD, Lloyd D, Tracey I, Vogt BA, Nadeau V,Jones AK. Placebo conditioning and placebo analgesia modulate a commonbrain network during pain anticipation and perception. Pain 2009;145:24–30.

[43] Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike MA. Maskedpresentations of emotional facial expressions modulate amygdala activitywithout explicit knowledge. J Neurosci 1998;18:411–8.

[44] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. ActaPsychiatr Scand 1983;67:361–70.

[45] Zubieta JK, Stohler CS. Neurobiological mechanisms of placebo responses. AnnN Y Acad Sci 2009;1156:198–210.