P

r

GASTROENTEROLOGY 2011;141:909–917

BASIC AND TRANSLATIONAL—ALIMENTARYTRACT

Neuroticism Influences Brain Activity During the Experience of Visceral PainSTEVEN J. COEN,* MICHIKO KANO,*,‡ ADAM D. FARMER,* VEENA KUMARI,§ VINCENT GIAMPIETRO,�

MICK BRAMMER,� STEVEN C. R. WILLIAMS,� and QASIM AZIZ*

*Wingate Institute of Neurogastroenterology, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London, England;‡ § �

Behavioral Medicine, Tohoku University, Sendai, Japan; and Departments of Psychology and Neuroimaging, Centre for Neuroimaging Sciences, Institute of

sychiatry, King’s College London, London, England

a

h

danpie

BA

SIC

AN

DTR

AN

SLA

TIO

NA

LA

T

BACKGROUND & AIMS: One particularly importantindividual dynamic known to influence the experience ofpain is neuroticism, of which little is known about invisceral pain research. Our aim was to study the relation-ship between neuroticism, psychophysiologic response, andbrain processing of visceral pain. METHODS: Thirty-onehealthy volunteers (15 male; age range, 22–38 years) partic-ipated in the study. The Eysenck Personality Question-naire was used to assess neuroticism. Skin conductancelevel, pain ratings, and functional magnetic resonanceimaging data were acquired during anticipation of painand painful esophageal distention. The effect of neuroti-cism was assessed using correlation analysis. RESULTS:There was a wide spread of neuroticism scores (range,0 –22) but no influence of neuroticism on skin conduc-tance level and pain tolerance or pain ratings. However, apositive correlation between brain activity and neuroti-cism during anticipation was found in regions associatedwith emotional and cognitive pain processing, includingthe parahippocampus, insula, thalamus, and anterior cingu-late cortex. These regions showed a negative correlation withneuroticism during pain (P � .001). CONCLUSIONS: Thisstudy provides novel data suggesting higher neuroticism isassociated with engagement of brain regions responsiblefor emotional and cognitive appraisal during anticipationof pain but reduced activity in these regions during pain.This may reflect a maladaptive mechanism in those withhigher neuroticism that promotes overarousal during an-ticipation and avoidance coping during pain.

Keywords: Pain; fMRI; Human Brain; Neuroticism.

Knowledge of human brain-gut interactions in healthand disease, with particular reference to visceral pain,

has progressed substantially since the advent of neuroim-aging techniques such as functional magnetic resonanceimaging (fMRI) and positron emission tomography. Earlystudies focused on addressing the sensory component ofvisceral pain by studying viscerotropic representation ofpainful and nonpainful visceral stimulation as well as inves-tigating the relationship between intensity of pain and brainactivity.1,2 Given the psychological nature of the pain expe-

ience, the studies that followed focused on how cognitive

nd affective factors modulate pain.3–6 To date, these inves-tigations have used groups of volunteers. However, there isstill a paucity of data on how individual differences modu-late perceptual and brain responses to visceral pain. Thisknowledge gap was highlighted in a recent Rome workingteam report that cited interindividual variability as a keylimiting factor in the development of brain imaging as a toolor biomarker for studying visceral pain.7

One particularly important interindividual factor thatis known to affect response to negative emotional stimuli,such as visceral pain, is the personality trait of neuroti-cism. Neuroticism is defined as a tendency to experiencenegative emotions, particularly in stressful situations.8

High neuroticism is associated with increased negativeaffect and anxiety, as well as with increased autonomicand neural responses to negative stimuli.9,10 Importantly,neuroticism has consistently been implicated as an impor-tant risk factor in the development of chronic somatic andvisceral pain following injury11 or surgery.12 Neuroticism is

igher in patients with irritable bowel syndrome (IBS)13 andis a risk factor for chronic unexplained pain in IBS.14 Recent

ata have also emerged that show higher neuroticism isssociated with decreased pain tolerance and a predomi-ately parasympathetic brainstem-mediated activity duringainful visceral stimulation.15 Others have shown that the

nfluence of neuroticism is not simply limited to the painfulvent but also to the anticipation of pain.16

Eysenck suggested the neurobiological substrate ofneuroticism is localized to the limbic system and hypo-thalamus, or the “the visceral brain system”; several ofthese regions are involved in pain processing.17 Despitethis, relatively few neuroimaging studies have probed therelationship between neuroticism, pain, and the brain.Using electroencephalography, Pauli et al showed thatpain sensitivity is associated with increased right hemi-sphere activity and with high neuroticism.18 Vossen et al

Abbreviations used in this paper: ACC, anterior cingulate cortex; fMRI,functional magnetic resonance imaging; IBS, irritable bowel syndrome;PCC, posterior cingulate cortex; PTT, pain toleration threshold; SCL, skinconductance level; VAS, visual analog scale.

© 2011 by the AGA Institute0016-5085/$36.00

doi:10.1053/j.gastro.2011.06.008

ipar

npadhbpt

ru(

Gb

BA

SICA

ND

TRA

NSLA

TION

AL

AT

910 COEN ET AL GASTROENTEROLOGY Vol. 141, No. 3

showed that higher neuroticism is associated with higheramplitudes of event-related cortical potentials to electricalpain.19 The influence of neuroticism on brain response tothe threat of pain (anticipatory fear) has also been inves-tigated using fMRI by Kumari et al, who found neuroti-cism correlated negatively with brain activity in manyregions known to be involved in the processing of visceralpain, including the thalamus, anterior cingulate cortex(ACC), and hippocampus.20 Importantly, the brain imag-ng studies listed previously investigated only somaticain or threat of somatic pain, highlighting the need formore extensive study on the relationship between neu-

oticism and visceral pain.In summary, there is a plethora of evidence to suggest

euroticism has a significant influence on the visceralain experience and that differences in neuroticism mayffect, or be driven by, the brain response and processinguring anticipation and pain itself. However, as yet thereas not been a study to assess the extent to which therain response to visceral pain and anticipation of visceralain is influenced by neuroticism. Our aim was to addresshis knowledge gap.

Subjects and MethodsVolunteersThirty-one healthy volunteers (15 male; mean age, 30

years; range, 22–38 years; all right-handed) participated in thestudy after providing informed written consent and approvalwas obtained from the local ethics committee (reference CREC/07/08-7). All volunteers were screened for a history of psychiatricor gastrointestinal symptoms and were not taking any medica-tions. Subjects completed the Eysenck Personality Question-naire-Revised (EPQ-R)21 to assess neuroticism level (neuroticismange, 0 –24; higher scores represent higher neuroticism). Vol-nteers also completed the Spielberger State and Trait Anxiety

STAI) Questionnaire22 to assess state (anxiety level on the day ofscanning) and trait anxiety (STAI range, 20 – 80; higher scoresrepresent higher anxiety). The questionnaires used are notscreening tools but measures of a spectrum of anxiety andneuroticism traits, and therefore extremely high or low scoresare not indicative of pathology. Both questionnaires were ad-ministered immediately before intubation and scanning.

Esophageal StimulationDuring fMRI scanning, painful phasic esophageal stim-

ulation (1-second duration) was delivered at pain tolerationthreshold by mechanical stimulation (balloon distention) of thedistal esophagus as previously described.4 The onset timing ofeach stimulus was achieved using customized computer soft-ware that was synchronized with scanner acquisition. Followingintubation (per-nasal), the catheter (commercially availableesophageal distention catheter; Sandhill Scientific, Oxford, Eng-land) was positioned so that the center of the balloon was in thedistal esophagus (35 cm from the nose). Sensory threshold andpain toleration threshold (PTT) was measured in each volunteer.To determine sensory threshold and PTT, the volume of disten-tion was increased in steps of 2-mL increments (from zero) untilthe volunteers first felt a sensation in the esophagus (sensorythreshold) and then until the point at which they could no

longer tolerate an increase in stimulus (PTT). The PTT level was P

measured for each volunteer immediately before positioning inthe MRI scanner and used as the level of stimulation throughoutthe functional brain imaging scan.

Experimental TaskAn event-related design was used. To measure anticipa-

tion of pain, we used visual cues (delivered by a program devel-oped in conjunction with Jeff Dalton, Centre for NeuroimagingSciences, Institute of Psychiatry, King’s College London), thedetails of which were described to volunteers before scanning.The functional imaging scan consisted of 20 trials incorporating60 events: 20 periods of anticipation, 20 painful esophagealdistentions, and 20 null events (rest/baseline). Each trial com-menced with the presentation of a visual warning cue (a yellowsquare) signaling a painful visceral stimulus was imminent (an-ticipation). The next event was a painful visceral stimulus lasting1 second (pain), which was immediately followed by a secondvisual cue (a blue square) signaling a “safe” condition duringwhich there was no risk of stimulation. Anticipation alwayspreceded pain, but the timing (onset) of each condition (antic-ipation and pain) was pseudo-randomized and jittered to therepetition time (TR); the anticipation phase signaled the start ofa new trial and lasted between 3 and 12 seconds, and the onsetof pain therefore was randomized to commence between 3 and12 seconds after the onset of a new trial while the safe periodlasted between 28 and 35 seconds following pain. Because thiswas an event-related design, we did not model the entire safeperiod as the rest or control condition. Instead, the middleperiod of the safe phase included a null event (subjects viewingthe rest cue) that served as the baseline or rest condition towhich other conditions were compared. The onset of the nullevent was jittered and modeled as a 1-second event in the sameway as the active conditions. The use of colored squares as cueswas counterbalanced to avoid any possible confounding effectsof color plus condition (pain or anticipation) pairing, such thathalf the volunteers received the blue square as a warning signaland yellow as safe while the others received the yellow square asa warning and blue as safe. The variable-onset timing of bothexperimental conditions (anticipation and pain) was pseudo-randomized as described previously to increase the unpredict-ability of stimuli, thereby reducing the effects of habituationand increasing the efficacy of the anticipation period.23

Subjective Perception of PainA visual analog scale (VAS; 5-second duration) measuring

the subjective perception of the stimulus (0 � no sensation, 50 �discomfort, 100 � extreme pain) was randomized to appear be-tween 9 and 15 seconds after each esophageal stimulation.

Skin Conductance LevelSkin conductance level (SCL) during anticipation, pain,

and rest was measured throughout the experimental task byattaching silver/silver chloride electrodes to the middle phalangeof the middle and index fingers of the left hand (SCL measuredin microsiemens using a constant voltage method [0.5 V], signaldigitized at 25-millisecond intervals [Contact Precision Instru-ments, Boston, MA]).

fMRIfMRI data (T2*-weighted images) were collected on a

eneral Electric (Milwaukee, WI) Signa Excite II 3.0 T scannerased at the Centre for Neuroimaging Sciences, Institute of

sychiatry, King’s College London. Head movement was re-

3m

mIptsf

as

caardawpaug

ui

tP

BA

SIC

AN

DTR

AN

SLA

TIO

NA

LA

T

September 2011 NEUROTICISM AND VISCERAL PAIN 911

stricted to a minimum by the use of foam padding within thehead coil. While inside the scanner, participants could view ascreen on which an electronic version of a VAS and the cue-related colored squares were projected. To record subjectiveratings of visceral sensation via VAS, an MRI-compatible buttonbox was placed in the right hand of each participant so theycould respond when the VAS was presented.

Before the start of the fMRI experiment, a high-resolutiongradient echo structural scan (43 � 3–mm slices, 0.3 interslicegap, echo time (TE) 30 milliseconds, TR 3000 milliseconds, flipangle 90°, matrix 1282, in-plane voxel dimensions 1.875 �1.875) was acquired to be used for Talairach normalization.During fMRI, a total of 480 T2*-weighted images per slice (40 �

–mm slices, 0.3 interslice gap, TE 30 milliseconds, TR 2500illiseconds, flip angle 80°, matrix 642, total number of images

per scan 19,200) depicting blood oxygen level– dependent con-trast were collected while participants underwent the experimen-tal task.

Statistical Analysis of fMRI DataAll MRI data were analyzed using XBAM (http://brain-

ap.co.uk/), an fMRI analysis software package developed at thenstitute of Psychiatry, King’s College London, which implementsermutation-based methods to minimize the number of assump-ions used in making statistical inference. fMRI data preprocessing,moothing, and individual brain activation mapping were per-ormed according to the method previously described.4

Correlation AnalysisIn the first instance, correlation analysis was performed

between the blood oxygen level– dependent effect data for eachindividual and their neuroticism score. This proceeds by firstchoosing the statistical map (fMRI response) corresponding tothe particular experimental contrast of interest. The Pearsonproduct moment correlation coefficient is then computed be-tween the neuroticism score for each subject and the fMRIresponses at each voxel for each subject, yielding one correlationcoefficient (r) per intracerebral voxel. To determine the signifi-cance of these correlation coefficients, the appropriate null dis-tribution of r is computed robustly using data permutation. Theorder of the behavioral data is randomly permuted withoutreplacement (ie, each data value occurs once, but the order ischanged), breaking the association between individual behav-ioral data and their corresponding fMRI responses. The corre-lation coefficient is then recomputed many times at each voxeland the resulting values of r combined over all voxels to produce

whole brain null distribution of r. The critical value of r forignificance at any particular P value can then be obtained from

this distribution after simply sorting it by value of r and select-ing the appropriate point from the sorted distribution; forexample, the critical value of r for a one-tailed test at a P valueof .05 would be the value of r in the null distribution chosensuch that 95% of all the null values of r lay below that point.Testing can then be extended to cluster level as described previ-ously. The cluster probability under the null hypothesis can bechosen to set the level of expected type I error clusters at anacceptable level (eg, �1 per whole brain).

Analysis of Psychophysical DataSCL amplitude to each stimulus event was analyzed

using a custom-made software program, SC-ANALYZE (Neuro-imaging Research, Institute of Psychiatry, London, England).

SCL (in microsiemens) was defined as the maximum increase 1 w

to 4 seconds after onset of condition and therefore representsthe maximum arousal to each stimulus. This time window isvariable due to the uncertain characteristics of SCL response andhas been previously validated.24,25 Mean SCL level for eachondition was calculated per individual and then averagedcross the group to produce mean SCL for anticipation, pain,nd rest. Repeated-measures analysis of variance with Bonfer-oni correction was used to examine differences in mean SCLuring rest, anticipation, and pain. All correlation analysis, suchs examining the relationship between neuroticism and PTT,as performed using Pearson’s correlation coefficient test andartial correlation where necessary (ie, to correct for the effect ofnxiety on brain responses). All behavioral data were processedsing SPSS (http://www.spss.com/) and GraphPad Prism (www.raphpad.com).

ResultsGroup Psychometric DataAll 31 volunteers tolerated the study well. There

was a wide range of neuroticism scores (mean, 8.1 � 1.2;range, 0 –22) with no significant difference between maleand female subjects (mean neuroticism � SEM; male, 9 �1.7 [range, 0 –20]; female, 7.4 � 1.4 [range, 1–20]; P � .49,t � 0.68, df 29). There was also a scope of state/trait anxietyscores across the group (mean state anxiety, 29.7 � 1.2[range, 20 – 41]; mean trait anxiety, 34.4 � 1.7 [range,20 –58]).

Group Physiologic DataAll volunteers rated the esophageal stimulus as pain-

ful (mean VAS � SEM, 63 � 1.9). The mean balloon volume(in milliliters) to reach sensory threshold was 7.6 � 0.9 andpain toleration threshold was 22 � 1.2. Mean group SCLincreased progressively from baseline to pain (mean SCL �SEM at baseline 2.4 � 0.39, anticipation 2.6 � 0.4, and pain2.9 � 0.4; P � .0001, F � 70.7, R2 � 0.72). Post hoc testsshowed that mean SCL during anticipation was significantlyhigher than baseline (P � .001) and SCL during pain wassignificantly higher than anticipation (P � .001).

Group Brain ActivityBrain activation during anticipation and painful

stimulation. The neural correlates of painful visceral stim-lation and anticipation have previously been well described

n the literature26; as such, the results of the present studywill focus on the influence of neuroticism on brain responsesduring these conditions. A summary of brain regions show-ing a significant increase in activity during pain and antici-pation can be found in Supplementary Table 1.

Effect of NeuroticismRelationship between neuroticism and psycho-

physiologic response. There was no significant correla-ion between neuroticism and pain ratings (r � �0.18,� .32), pain tolerance volume (r � �0.18, P � .31), or

SCL response during pain (r � 0.003, P � .98) andanticipation of pain (r � 0.01, P � .93). However, there

as a positive correlation between neuroticism and state

a

rnibtwrSc

aisccppsF

.

tg

2211111

C544211111

NrsB

BA

SICA

ND

TRA

NSLA

TION

AL

AT

912 COEN ET AL GASTROENTEROLOGY Vol. 141, No. 3

(r � 0.4516, P � .01) and trait (r � 0.6385, P � .001)nxiety.

Relationship between neuroticism and brain activ-ity during anticipation of pain. A significant positive cor-elation between brain activity during anticipation andeuroticism score was found in several brain regions,

ncluding the right parahippocampal gyrus, right insula,ilateral thalamus, and ACC (BA32). In contrast, a nega-ive correlation between neuroticism and brain activityas evident in regions including bilateral superior tempo-

al gyrus, precuneus, and posterior cingulate cortex (PCC).ee Table 1 for a summary of all brain regions showingorrelations and Figure 1 for representative plots.

Relationship between neuroticism and brain activ-ity during pain. During pain, there was a positive associ-

tion between level of brain activity and neuroticism scoren the right hemisphere in the insula, amygdala, primaryomatosensory cortex (SI), and left hemisphere in theulmen and middle temporal gyrus. There was a negativeorrelation in several areas, including bilateral thalamus,arahippocampal gyrus and insula, right ACC (BA24), leftutamen, and ventral ACC (BA32). See Table 2 for aummary of all brain regions showing correlations andigure 1 for representative plots.

Effect of anxiety. Because there was a relationshipbetween anxiety and neuroticism, a partial correlationanalysis was conducted to determine the relationship be-tween neuroticism and brain activity when corrected foranxiety. This analysis showed that the correlation between

Table 1. Summary of Brain Regions Showing a Correlation BeVisceral Pain

Size P value R value X Y

Clusters showing a positive correlation between brain activity during a55 .00004 0.48 22 1736 .00004 0.38 8 2727 .00004 0.52 �48 �2925 .00004 0.47 11 �194 .00004 0.39 41 80 .00004 0.38 �34 67 .00004 0.42 17 -35 .00004 0.50 �7 414 .00008 0.34 4 192 .00004 0.35 7 �631 .00004 0.40 �6 �7

lusters showing a negative correlation between brain activity during3 .00004 0.44 �56 �214 .00004 0.38 �37 �552 .00004 0.46 43 �559 .00004 0.48 �12 �398 .00004 0.45 49 �388 .00004 0.49 26 �665 .00004 0.41 6 �583 .00004 0.41 �41 �613 .00012 0.33 �31 28

OTE. Talairach and Tournoux coordinates are expressed in millimetersepresent points of maximum activation at the group level (highest mtatistics and therefore do not have cluster size limitations.A, Brodmann area.

neuroticism and brain activity is indeed mediated by neu-

roticism (ie, the correlations remained significant whenanxiety was accounted for) with the exception of thepositive correlation between neuroticism and brain activ-ity in the cerebellum during anticipation (r � 0.26, P �15), the SI (r � 0.34, P � .06) and middle temporal gyrus(r � 0.29, P � .12) during pain, and the negative correla-ion with neuroticism during pain in the superior frontalyrus (r � �0.26, P � .16).

DiscussionTo our knowledge, this is the first study to report

a significant effect of neuroticism on brain processingduring visceral pain stimulation and anticipation. Thelevel of brain activity during anticipation and pain varieddepending on neuroticism score in several brain regionspreviously shown to be involved in pain processing, in-cluding the mid-ACC (BA24), perigenual ACC (BA32),insula, thalamus, inferior frontal gyrus, PCC, and supple-mentary motor area (SMA).

Behavioral FindingsThere was an increase in SCL during anticipation

compared with baseline. This reflects an increase in sym-pathetic activity during anticipation compared with base-line (safe period) and provides evidence that the study designwas effective in eliciting an anticipatory response becausevolunteers showed increased autonomic arousal during thisperiod. When the influence of neuroticism was examined, no

en Neuroticism and Brain Activity During Anticipation of

Z Side Brain region

cipation and neuroticism scores�5 Right Putamen24 Right ACC (BA32)24 Left Secondary association area (SII)7 Right Thalamus

�3 Right Anterior insula�1 Left Claustrum

�11 Right Parahippocampal gyrus20 Left Medial frontal gyrus (BA9)43 Right Medial frontal gyrus (BA8)

�40 Right Cerebellum (posterior)0 Left Thalamus

icipation and neuroticism scores�3 Left Middle temporal gyrus (BA21)14 Left Superior temporal gyrus (BA22)14 Right Superior temporal gyrus (BA22)37 Left Cingulate cortex (BA31)

�11 Right Fusiform gyrus (BA37)34 Right Precuneus8 Right PCC (BA30)

37 Left Angular gyrus38 Left Middle frontal gyrus (BA8)

y, z); size indicates number of voxels. The coordinates for each clusteran response in the cluster). Clusters are defined using cluster mass

twe

nti

ant

(x,edi

significant correlation between SCL and neuroticism was

BA

SIC

AN

DTR

AN

SLA

TIO

NA

LA

T

September 2011 NEUROTICISM AND VISCERAL PAIN 913

Figure 1. Representative graphical examples of correlations (positive, red; negative, blue) found between neuroticism and level of brain activity(SSQ) during (A) anticipation and (B) pain. This figure shows several brain regions, including the thalamus, parahippocampal gyrus, and ACC, where

brain activity increases with higher neuroticism during anticipation but decreases during pain processing.

rmacsa

3111

C7633221111

NrsB

BA

SICA

ND

TRA

NSLA

TION

AL

AT

914 COEN ET AL GASTROENTEROLOGY Vol. 141, No. 3

found. The fact that a significant effect of neuroticism onsympathetic nervous system response during pain and an-ticipation was not found is not entirely surprising becauseseveral studies have shown mixed results, with some showingincreased SCLs in individuals with high neuroticism andothers showing the opposite finding.15,27,28

Analysis of the psychophysical data provided no evi-dence of an effect of neuroticism on threshold balloonvolumes or pain ratings. This finding is consistent withprevious studies suggesting that sensory mechanisms ofnociceptive processing such as pain scores or thresholdsare not related to neuroticism. Rather, it is the cognitiveand emotional aspects of pain processing such as suffer-ing/unpleasantness, affective disturbance related topain,29 and pain-related fear during vigilance to pain thatare affected by neuroticism.30 In particular, Harkins et al29

showed that higher neuroticism was associated with sig-nificantly higher unpleasantness ratings. We did not ac-quire unpleasantness ratings in the present study becausewe did not want to compromise the brain response byhaving too many tasks while the volunteers were scanned.Pain intensity ratings were chosen in preference because itwas important to confirm the stimuli were indeed painfulfor each volunteer throughout the scanning session. Al-though in our study neuroticism did not affect painresponses, it is important to acknowledge that the volunteersample used, although more than adequate for an fMRIstudy, may not be large enough to detect subtle changes insubjective behavioral responses such as pain ratings. Indeed,previous work involving a larger sample of healthy volun-teers has shown lower pain toleration thresholds to visceralpain in those with higher neuroticism.31

Investigation of the psychometric data revealed a posi-tive correlation between neuroticism and anxiety, which

Table 2. Summary of Brain Regions Showing a Correlation Be

Size P value R value X Y

Clusters showing a positive correlation between brain activity during p44 .00004 0.42 �11 �51 .00004 0.42 40 �25 .00004 0.42 24 �2 .00004 0.41 43 �10 .00004 0.34 �46 �5

lusters showing a negative correlation between brain activity during6 .000041 0.55 4 37 .000041 0.47 14 �28 .000041 0.44 �45 �26 .000041 0.51 19 �34 .000041 0.40 �21 �21 .000041 0.35 �11 �28 .000041 0.42 �16 �7 .000041 0.37 33 21 .000041 0.32 �8 30 .000041 0.37 �31 �2

OTE. Talairach and Tournoux coordinates are expressed in millimetersepresent points of maximum activation at the group level (highest mtatistics and therefore do not have cluster size limitations.A, Brodmann area.

was not an unexpected finding. It is difficult to disentan-

gle the effect of anxiety from neuroticism because in-creased negative affect such as high anxiety is one of thefactors that contribute to neuroticism as a personalitydimension.8,32 Furthermore, evidence suggests that neu-oticism and anxiety are correlated and refer to approxi-

ately the same emotional condition.33 Indeed, Cattellnd Scheier, who devised their own anxiety measure, con-luded that Eysenck’s neuroticism dimension (as mea-ured in the present study) is identical to their concept ofnxiety.34 Nevertheless, a question can be raised about

whether it is neuroticism or anxiety mediating the brainresponse during anticipation and pain in the currentstudy. However, it is well known that high anxiety in-creases SCL, reduces pain tolerance and sensory thresh-olds, and exacerbates pain perception. In the presentstudy, there was no relationship between anxiety and painscores or stimulation thresholds. This suggests that thelevel of anxiety, although greater in volunteers with highneuroticism, was not significant enough to influence thebehavioral and cerebral response during pain and antici-pation of pain. This is supported by the fact that all brainregions remained significantly correlated with neuroti-cism even when the effect of anxiety was controlled for,with the exception of the cerebellum during anticipationand postcentral gyrus, middle temporal gyrus, and supe-rior frontal gyrus during pain. This suggests that al-though anxiety is an important factor, the relationshipbetween brain activity and neuroticism during pain andanticipation found in the present study is independent ofanxiety.

Brain Activity During AnticipationAnticipation of pain resulted in increased brain

activity in several regions, including the anterior insula,

en Neuroticism and Brain Activity During Visceral Pain

Z Side Brain region

and neuroticism scores�16 Left Culmen

16 Right Posterior insula�18 Right Amygdala

46 Right Postcentral gyrus2 Left Middle temporal gyrus

and neuroticism scores13 Right ACC (BA24)10 Right Thalamus21 Left Mid insula�1 Right Parahippocampal gyrus�7 Left Parahippocampal gyrus10 Left Thalamus65 Right Superior frontal gyrus (BA6)13 Right Anterior insula13 Left ACC (BA32)2 Left Putamen

y, z); size indicates number of voxels. The coordinates for each clusteran response in the cluster). Clusters are defined using cluster mass

twe

ain26999

pain3623523361

(x,edi

ACC (BA24 and 32), frontal cortex (BA9 and 10), and

hicnttr

bt(cpebwio

t

pbtdtcsb

ait

dtriisratoahw

BA

SIC

AN

DTR

AN

SLA

TIO

NA

LA

T

September 2011 NEUROTICISM AND VISCERAL PAIN 915

secondary somatosensory cortex, which is consistent withprevious studies assessing anticipation of visceral pain.6,35

Neuroticism and Brain Activity DuringAnticipationDuring anticipation, there was a negative correla-

tion between brain activity and neuroticism in the middleand superior temporal gyrus, precuneus, and PCC, con-sistent with the findings of Kumari et al, who foundnegative correlations in the same brain regions duringanticipatory fear induced by the threat of pain.20 The PCC

as been implicated in encoding stimulus intensity36 dur-ng pain. However, the present study showed no signifi-ant effect of neuroticism on the sensory discriminativeature of the stimulus (pain scores, thresholds), andherefore neuroticism-associated differences in brain ac-ivity in the PCC in the present study are unlikely to beelated to encoding stimulus intensity.

In the present study, we found a positive relationshipetween brain activity and neuroticism during anticipa-ion in several regions. These regions include the ACCBA32), parahippocampal gyrus, anterior insula, frontalortex, putamen, and bilateral thalamus, all of which arerimarily believed to be involved in the cognitive andmotional modulation of pain. The positive relationshipetween brain activity and neuroticism in the thalamusas localized to the medial dorsal nucleus, which has been

mplicated in attention, preparation/planning, and mem-ry during fear anticipation.37 This area has connections

with the frontal cortex, which also showed a positivecorrelation and has been widely implicated in cognitivefunctioning, including that during visceral pain.3

There was a positive correlation with neuroticism in theACC (BA32) and brain activity during anticipation. Sev-eral neuroimaging studies have associated the ACC(BA32) with the neural basis for mood disorders,38 nega-ive emotional states,39 and negative affect in pain4,5,40;

therefore, higher ACC activity may reflect the tendency forthose with higher neuroticism to experience negativeemotions during stressful situations.8 It should be notedthat this finding is not consistent with that of Kumari etal, who found a negative correlation in this region. How-ever, Kumari et al actually found a negative correlation inthe left hemisphere whereas the positive correlation in thepresent study was in the right hemisphere; this may reflectspecificity of brain response between somatic and visceralpain threat but is more likely due to several differences inmethodology between the studies.

When considering the results of the current study withthat of Kumari et al,20 it is important to note that the

resent study also showed positive correlations in severalrain regions, whereas no positive correlations were de-ected in the study by Kumari et al. The reason for thisifference is probably that the current study used doublehe number of volunteers, which could quite easily ac-ount for the increased statistical power of the presenttudy to detect significant positive correlations. The study

y Kumari et al also did not involve pain, so it is conceiv-

ble that the anticipation period was not as threatening ast could have been had volunteers received pain intermit-ently to reinforce the anticipation response.

Brain Activity During PainBrain activity during pain increased in a number of

brain regions previously implicated in visceral pain pro-cessing and commonly referred to as the visceral painmatrix.26 This included activity of the primary and sec-ondary somatosensory cortex, thalamus, ACC, amygdala,and anterior insula.

Neuroticism and Brain Activity During PainThere was a negative correlation between neuroti-

cism and brain activity during pain in several brain re-gions, including the ACC (BA32), bilateral thalamus, para-hippocampal gyrus, anterior and mid insula, andputamen. It is striking to note that the majority of theseregions are the same as those showing a positive correla-tion during anticipation.

The negative relationship between neuroticism andbrain activity in the thalamus was specifically localized tothe pulvinar nucleus, a region that is directly connected tothe ACC and amygdala and previously implicated in re-laying information related to the emotional componentof a stimulus to the limbic system.41 This suggests that

ifferences in brain activity linked to neuroticism in thehalamus during pain may be related to the emotionalesponse to the stimulus. The pattern of amygdala activ-ty, which was greater in those with higher neuroticism, isn contrast to several other regions of the limbic systemhowing a decrease in activity in those with higher neu-oticism during the pain period. Activation of themygdala with simultaneous suppression of other cogni-ive-emotional brain regions could be explained by previ-us studies suggesting that increased amygdala activity isssociated with passive coping strategies such as learnedelplessness,42 particularly when humans are presentedith an unsolvable cognitive problem.43 In the present

study, it is possible the amygdala response is involved insuppression of other brain regions because the unavoid-able painful stimulus evokes a greater state of learnedhelplessness in the volunteers with higher neuroticism.

Higher neuroticism was also associated with reducedbilateral parahippocampal gyrus activity during pain. Thisis particularly interesting because this region has beenwidely implicated in processes related to memory, andtherefore this relationship may reflect poorer memoryprocessing during pain in those with higher neuroticism.This suggestion is supported by several studies showingreduced memory performance during stress in those whoscore highly on neuroticism.44

Given the fact that neuroticism is considered a riskfactor for IBS, it is pertinent to note that a recent meta-analysis of brain activity in patients with IBS comparedwith controls suggests that regions involved in emotionalarousal (eg, amygdala) and homeostasis are more active in

patients.45 Many of these regions were also more active in

BA

SICA

ND

TRA

NSLA

TION

AL

AT

916 COEN ET AL GASTROENTEROLOGY Vol. 141, No. 3

volunteers with higher neuroticism during anticipation andthe amygdala during pain. In contrast, decreased activity inseveral of these regions during pain seems to be contradic-tory to that seen in patients with IBS. However, relating thepresent findings to neuroimaging studies of brain responseto pain in patients with IBS is somewhat problematic be-cause the majority of studies have not explicitly examinedthe role of anticipation and pain separately. The role ofanticipation should not be underestimated because previousstudies have shown that anticipation significantly contrib-utes to the brain response seen during pain6 and this may bethe case in brain imaging studies of IBS, as also suggested inthe meta-analysis by Tillisch et al.45

Taken together, the data provide considerable evidenceto suggest an overall trend toward greater activity in brainregions associated with cognitive-emotional processing inthose with higher neuroticism during anticipation ofpain, which may reflect increased cognitive and emotionalprocessing during threat of an aversive stimulus in thesevolunteers. In contrast, higher neuroticism is associatedwith a suppression of this response during pain. This is animportant finding and may reflect a tendency for thosewith higher neuroticism toward anticipatory anxiety anda lack of coping when the source of anxiety (such as pain)becomes a reality. This type of behavior has been noted inprevious studies that have suggested neuroticism in-creases vigilance for distressing/threatening stimuli butpromotes emotional blunting when escape is not an op-tion46 and avoidance/passive coping strategies.42,47 In ad-dition, the results show little support for neuroticismmodulation of brain regions believed to be involved insensory-discriminative responses to pain and thereforesupport those theories that suggest it is the cognitive andemotional aspects of pain processing that are affected byneuroticism29,30 and not the sensory aspects per se.

LimitationsA whole brain analysis approach to examine the

modulation of brain response to pain and anticipation byneuroticism was used. This exploratory approach wasused because the effect of neuroticism, although exam-ined to somatic pain, has not previously been assessedusing visceral pain. As such, using an a priori hypothesisbased on somatic pain risks excluding associations thatare specific to visceral pain. Nevertheless, the limitation ofthe whole brain analysis approach is reflected in the re-sults because although positive correlations during antic-ipation and negative correlations during pain were seen inthe same brain regions, they were not exactly the samecoordinates or voxels. There were, for example, varioussubregions of ACC and thalamus that were different be-tween anticipation and pain. This should not weaken theimportance of the present findings, which do show re-gions positively correlated during anticipation to be neg-atively correlated during pain. However, future researchcould adopt a region of interest approach to determine in

more detail the proximity of regions positively correlated

with neuroticism during anticipation that are negativelycorrelated during pain.

Summary and ConclusionThis study provides novel data suggesting that

higher neuroticism is associated with greater activity inlimbic areas of the brain during anticipation of visceralpain but with lower activity in the same brain regionsduring painful visceral stimulation. In the context ofneuroticism and pain, this is important because it mayreflect a maladaptive mechanism in those with higherneuroticism that promotes overarousal during anticipa-tion, which may in turn promote a lack of coping duringpain. This behavior may help explain the greater incidenceof those with higher neuroticism attending outpatientpain clinics and being at greater risk for developingchronic pain conditions. Furthermore, this theory findssupport in several studies investigating functional gastro-intestinal disorders associated with visceral pain (such asIBS) that have shown differences between patients andhealthy controls in several of the brain regions identifiedas being linked with neuroticism in our study.45

Finally, this investigation has shown that individualdifferences in neuroticism affect brain response duringthe experience of visceral pain and provides knowledge onone of the factors responsible for interindividual variabil-ity in brain responses during visceral pain. In doing so, theresults of this study also provide rationale for more de-tailed volunteer selection criteria with the aim of increas-ing homogeneity in groups used in experimental research.This approach may also help finesse the development ofnovel behavioral and pharmacologic interventions andimprove the utility of fMRI to identify biomarkers of pain.

Supplementary Material

Note: To access the supplementary materialaccompanying this article, visit the online version ofGastroenterology at www.gastrojournal.org, and at doi:10.1053/j.gastro.2011.

References

1. Aziz Q, Andersson JL, Valind S, et al. Identification of human brainloci processing esophageal sensation using positron emissiontomography. Gastroenterology 1997;113:50–59.

2. Aziz Q, Thompson DG, Ng VW, et al. Cortical processing of humansomatic and visceral sensation. J Neurosci 2000;20:2657–2663.

3. Coen SJ, Aziz Q, Yaguez L, et al. Effects of attention on visceralstimulus intensity encoding in the male human brain. Gastroen-terology 2008;135:2065–2074.

4. Coen SJ, Yaguez L, Aziz Q, et al. Negative mood affects brainprocessing of visceral sensation. Gastroenterology 2009;137:253–261.

5. Phillips ML, Gregory LJ, Cullen S, et al. The effect of negativeemotional context on neural and behavioural responses to oe-sophageal stimulation. Brain 2003;126:669–684.

6. Yaguez L, Coen S, Gregory LJ, et al. Brain response to visceralaversive conditioning: a functional magnetic resonance imagingstudy. Gastroenterology 2005;128:1819–1829.

7. Mayer EA, Aziz Q, Coen S, et al. Brain imaging approaches to thestudy of functional GI disorders: a Rome working team report.

Neurogastroenterol Motil 2009;21:579–596.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

BA

SIC

AN

DTR

AN

SLA

TIO

NA

LA

T

September 2011 NEUROTICISM AND VISCERAL PAIN 917

8. Costa PT, McCrae RR. Influence of extraversion and neuroticismon subjective well-being: Happy and unhappy people. J Pers SocPsychol 1980;38:668–678.

9. Norris CJ, Larsen JT, Cacioppo JT. Neuroticism is associated withlarger and more prolonged electrodermal responses to emotion-ally evocative pictures. Psychophysiology 2007;44:823–826.

0. Canli T. Functional brain mapping of extraversion and neuroticism:learning from individual differences in emotion processing. J Pers2004;72:1105–1132.

1. Taylor KS, Anastakis DJ, Davis KD. Chronic pain and sensorimotordeficits following peripheral nerve injury. Pain 2010;151:582–591.

2. Bisgaard T, Klarskov B, Rosenberg J, et al. Characteristics andprediction of early pain after laparoscopic cholecystectomy. Pain2001;90:261–269.

3. Hazlett-Stevens H, Craske MG, Mayer EA, et al. Prevalence ofirritable bowel syndrome among university students: the roles ofworry, neuroticism, anxiety sensitivity and visceral anxiety. J Psy-chosom Res 2003;55:501–505.

4. Gwee KA, Leong YL, Graham C, et al. The role of psychological andbiological factors in postinfective gut dysfunction. Gut 1999;44:400–406.

5. Paine P, Worthen SF, Gregory LJ, et al. Personality differencesaffect brainstem autonomic responses to visceral pain. Neurogas-troenterol Motil 2009;21:1155–e98.

6. Naliboff BD, Waters AM, Labus JS, et al. Increased acousticstartle responses in IBS patients during abdominal and nonab-dominal threat. Psychosom Med 2008;70:920–927.

7. Eysenck HJ. The biological basis of personality. Springfield, IL:Charles C Thomas, 1967.

8. Pauli P, Wiedemann G, Nickola M. Pain sensitivity, cerebral later-ality, and negative affect. Pain 1999;80:359–364.

9. Vossen HG, van Os J, Hermens H, et al. Evidence that trait-anxietyand trait-depression differentially moderate cortical processing ofpain. Clin J Pain 2006;22:725–729.

0. Kumari V, ffytche DH, Das M, et al. Neuroticism and brain responsesto anticipatory fear. Behav Neurosci 2007;121:643–652.

1. Eysenck HJ, Eysenck SBG. Manual of the Eysenck PersonalityScales. London, UK: Hodder and Stoughton, 1991.

2. Spielberger CD, Gorsuch RL, Lushene RE. Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting PsychologistsPress, 1970.

3. Carlsson K, Andersson J, Petrovic P, et al. Predictability modu-lates the affective and sensory-discriminative neural processing ofpain. Neuroimage 2006;32:1804–1814.

4. Williams AE, Rhudy JL. The influence of conditioned fear on humanpain thresholds: does preparedness play a role? J Pain 2007;8:598–606.

5. Kotses H, Glaus KD. Latency of multiple skin conductance responses indifferential classical conditioning. Biol Psychol 1977;5:1–6.

6. Van Oudenhove L, Coen SJ, Aziz Q. Functional brain imaging ofgastrointestinal sensation in health and disease. World J Gastro-enterol 2007;13:3438–3445.

7. Matthews G. Neuroticism from the top down: psychophysiologyand negative emotionality. In: Stelmack RM, ed. On the psy-chobiology of personality. Oxford, England: Elsevier, 2004:251–266.

8. Hagemann D, Naumann E, Lürken A, et al. EEG asymmetry, dis-positional mood and personality. Pers Individual Differences1999;27:541–568.

9. Harkins SW, Price DD, Braith J. Effects of extraversion and neu-roticism on experimental pain, clinical pain, and illness behavior.Pain 1989;36:209–218.

0. Goubert L, Crombez G, Van Damme S. The role of neuroticism,pain catastrophizing and pain-related fear in vigilance to pain: a

structural equations approach. Pain 2004;107:234–241.

1. Farmer AD, Coen SJ, Kano M, et al. Human psychophysiologicalresponses to visceral and somatic pain: towards an integratedreproducible pain endophenotype? Gut 2009;59:A137.

2. Gray JA, McNaughton N. The neuropsychology of anxiety. Oxford,UK: Oxford University Press, 2000.

3. Luteijn F, Bouman TK. The concepts of depression, anxiety, andneuroticism in questionnaires. Eur J Pers 1988;2:113–120.

4. Cattell RB, Scheier IH. The meaning and measurement of neurot-icism and anxiety. New York, NY: Ronald Press, 1961.

5. Berman SM, Naliboff BD, Suyenobu B, et al. Reduced brainsteminhibition during anticipated pelvic visceral pain correlates withenhanced brain response to the visceral stimulus in women withirritable bowel syndrome. J Neurosci 2008;28:349–359.

6. Tolle TR, Kaufmann T, Siessmeier T, et al. Region-specific encod-ing of sensory and affective components of pain in the humanbrain: a positron emission tomography correlation analysis. AnnNeurol 1999;45:40–47.

7. Li XB, Inoue T, Nakagawa S, et al. Effect of mediodorsal thalamicnucleus lesion on contextual fear conditioning in rats. Brain Res2004;1008:261–272.

8. Drevets WC, Price JL, Simpson JR Jr, et al. Subgenual prefrontalcortex abnormalities in mood disorders. Nature 1997;386:824–827.

9. Bishop S, Duncan J, Brett M, et al. Prefrontal cortical function andanxiety: controlling attention to threat-related stimuli. Nat Neuro-sci 2004;7:184–188.

0. Ochsner KN, Ludlow DH, Knierim K, et al. Neural correlates ofindividual differences in pain-related fear and anxiety. Pain 2006;120:69–77.

1. Ohman A. The role of the amygdala in human fear: automatic detec-tion of threat. Psychoneuroendocrinology 2005;30:953–958.

2. Amorapanth P, LeDoux JE, Nader K. Different lateral amygdalaoutputs mediate reactions and actions elicited by a fear-arousingstimulus. Nat Neurosci 2000;3:74–79.

3. Schneider F, Gur RE, Alavi A, et al. Cerebral blood flow changes inlimbic regions induced by unsolvable anagram tasks. Am J Psy-chiatry 1996;153:206–212.

4. Neupert SD, Mroczek DK, Spiro A. Neuroticism moderates thedaily relation between stressors and memory failures. PsycholAging 2008;23:287–296.

5. Tillisch K, Mayer EA, Labus JS. Quantitative meta-analysis identi-fies brain regions activated during rectal distension in irritablebowel syndrome. Gastroenterology 2011;140:91–100.

6. Wilson GD, Kumari V, Gray JA, et al. The role of neuroticism instartle reactions to fearful and disgusting stimuli. Pers IndividualDifferences 2000;29:1077–1082.

7. Endler NS, Parker JD. Multidimensional assessment of coping: acritical evaluation. J Pers Soc Psychol 1990;58:844–854.

Received January 24, 2011. Accepted June 3, 2011.

Reprint requestsAddress requests for reprints to: Steven J. Coen, PhD, Wingate

Institute of Neurogastroenterology, Barts and the London School ofMedicine and Dentistry, 26 Ashfield Street, London, E1 2AJ, England.e-mail: s.j.coen@qmul.ac.uk; fax: (44) 0 20 7375 2103.

Conflicts of interestThe authors disclose the following: Q.A. has received educational

grants from GlaxoSmithKline, Pfizer, and Novartis, none of which arerelevant to the work described in this report. S.J.C., M.K. A.D.F., V.K.,V.G., M.B., and S.C.R.W. disclose no conflicts.

FundingSupported jointly by a Medical Research Council grant (to Q.A.)

and a British Academy grant (to S.J.C.).

B21

917.e1 COEN ET AL GASTROENTEROLOGY Vol. 141, No. 3

Supplementary Table 1. Summary of Brain Regions Showing

Size X Y Z

rain activity during anticpation of painful visceral stimulation00 36 19 1089 29 �59 �20

164 �36 �25 53122 15 8 14117 22 �62 3389 �4 �59 4684 30 �73 �1181 25 �75 �361 47 19 360 �29 �52 �2649 �28 �82 247 �5 �20 246 �25 �70 1146 �4 11 3346 4 7 4045 4 0 4335 �43 11 328 7 �19 328 55 �23 1727 43 15 3023 �1 23 3116 52 �40 816 �27 43 2215 29 49 2111 �28 48 25

Brain activity during painful visceral stimulation346 �28 �11 47331 51 �7 7240 46 �8 37235 30 �2 2216 58 �11 10180 �44 �6 11179 23 1 �11160 56 �35 18146 �14 �56 �16140 11 �11 3138 �30 �2 2128 �51 �22 16127 55 �29 24116 6 0 3397 �7 �11 4096 4 �67 �1392 �38 �18 3838 13 �5 6429 �23 13 �1529 �22 �4 �2020 26 45 3120 5 �69 2113 45 31 2113 �7 �70 2310 �25 43 25

NOTE. Talairach and Tournoux coordinates in millimeters (x, y, z); sizpoints of maximum activation at the group level (highest median restherefore do not have cluster size limitations. All clusters corrected to

an Increase in Activity During Anticipation and Visceral Pain

Side Cerebral region

Right Anterior insulaRight Cerebellum (posterior)Left Precentral gyrus (motor cortex)Right CaudateRight PrecuneusLeft PrecuneusRight Fusiform gyrusRight Lingual gyrus (occipital lobe)Right Inferior frontal gyrus (BA45)Left Cerebellum (culmen)Left Lingual gyrus (occipital lobe)Left ThalamusLeft PCC (BA30)Left ACC (BA24)Right ACC (BA32)Right ACC (BA24)Left Anterior insulaRight ThalamusRight Secondary somatosensory cortex (SII)Right Middle frontal gyrus (BA9)Left ACC (BA32)Right Superior temporal gyrus (BA22)Left Middle frontal gyrus (BA10)Right Middle frontal gyrus (BA10)Left Superior temporal gyrus (BA22)

Left Middle frontal gyrus (BA6)Right Primary somatosensory cortex (SI)Right Precentral gyrus (motor cortex [BA6])Right PutamenLeft Primary somatosensory cortex (SI)Left Mid insulaRight Parrahippocampal gyrusRight Mid insulaLeft Cerebellum (culmen)Right ThalamusLeft PutamenLeft Secondary somatosensory cortex (SII)Right Secondary somatosensory cortex (SII)Right ACC (BA24)Left ACC (BA24)Right CerebellumLeft Precentral gyrus (motor cortex [BA6])Right Superior frontal gyrus (BA6)Left Inferior frontal gyrus (BA47)Left AmygdalaRight Superior frontal gyrus (BA9)Right PrecuneusRight Middle frontal gyrus (BA46)Left PrecuneusLeft Superior frontal gyrus (BA10)

e indicates number of voxels. The coordinates for each cluster representponse in the cluster). Clusters defined using cluster mass statistics andP � .003, ensuring a rate of less than one false-positive cluster per brain.

BA, Brodmann area.