A Prefrontal Non-Opioid Mechanism in Placebo Analgesia

 



PAIN 150 (2010) 59–65

www.elsevier.com/locate/pain

A prefrontal non-opioid mechanism in placebo analgesia Predrag Petrovic a, , Eija Kalso b, Karl Magnus Petersson a,c, Jesper Andersson d, Peter Fransson a, Martin Ingvar a *

a

Cognitive Neurophysiology Research Group, Stockholm Brain Institute, Osher Center for Integrative Medicine, Karolinska Institutet, Stockholm, Sweden Pain Clinic, Department of Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital and Institute of Clinical Medicine, University of Helsinki, Finland c Cognitive Neuroscience Research Group, IBB/CBME, University of Algarve, Faro, Portugal d Oxford Centre for Functional MRI of the Brain, University of Oxford, Oxford, UK  b

a r t i c l e

i n f o

 Article history: Received 23 August 2008 Received in revised form 17 February 2010 Accepted 10 March 2010

Keywords: Placebo Pain Opioid Cognitive Orbitofrontal cortex, prefrontal cortex, anterior cingulate cortex PET fMRI, expectation, error signal

a b s t r a c t

Behavior Beha vioral al studi studies es havesuggested havesuggested that placebo placebo ana analges lgesia ia is partly partly med mediate iated d by the endo endogeno genous us opioid opioid system. Expa tem. Expandin nding g on these these resu results lts wehave shown shown thatthe opio opioid-r id-recep eceptor-r tor-rich ich rostral rostral anteriorcingulat anteriorcingulate e cortex (rACC) is activated in both placebo and opioid analgesia. analgesia. However, there are also differences differences between between the two treatments. While opioids have direct pharmacological effects, acting on the descending pain inhibitory inhib itory syste system, m, plac placebo ebo analgesi analgesia a depends depends on neocorti neocortical cal top-down top-down mechanism mechanisms. s. An importa important nt differdifferen ence ce ma may y be that that expect expectati ations ons ar are e me mett to a les lesser ser ex exten tentt in pla placeb cebo o tre treatm atment ent as compa comparedwith redwith a spe specifi cific c tre treatm atment ent,, yieldi yielding ng a largererro largererrorr signal signal.. As these these proce processe ssess previ previous ously ly have have been been sho shown wn to infl influe uenceother nceother types of perceptual experiences, we hypothesized that they also may drive placebo analgesia. Imaging studies suggest that lateral orbitofrontal cortex (lObfc) and ventrolateral prefrontal cortex (vlPFC) are involvedin invo lvedin processin processing g expectati expectation on and erro errorr signals. signals. We re-analy re-analyzed zed two inde indepen pendent dent functiona functionall imaging imaging experiments related to placebo analgesia and emotional placebo to probe for a differential processing in these regions during placebo treatment vs. opioid treatment and to test if this activity is associated with the placeb placebo o respo response nse.. In the firs firstt datas dataset et lObfc lObfc and vlP vlPFC FC sho showe wed d an enhan enhanced ced activa activatio tion n in pla placeb cebo o an analalgesia gesi a vs. o opioi pioid d analgesi analgesia. a. Further Furthermore more,, the rACC acti activity vity co-v co-varie aried d with the prefronta prefrontall regions regions in the placebo condition specifically. A similar correlation between rACC and vlPFC was reproduced in another datas dataset et involv involving ing em emoti otiona onall pla placeb cebo o andcorrelat andcorrelated ed with with the degre degree e of the pla placeb cebo o eff effect ect.. Our result resultss thus thus support that placebo is different different from specific treatment treatment with a prefrontal top-do top-down wn influence on rACC.   2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction It is widely accepted that placebo analgesia is mediated by the endogen end ogenous ous opioid opioid syste system m sinc since e the opioid opioid rece recepto ptorr antagoni antagonist st naloxone attenuates the placebo effect  effect   [2,5–7,26]. [2,5–7,26]. In line with this we ha have ve shown shown a sim simila ilarr functi functiona onall anatom anatomy y of op opio ioid id and pla placeb cebo o [34].. Opi Opioid oid-ric -rich h regions regions includin including g the rost rostral ral ante anterior rior analgesia   [34] analgesia cingulate cortex (rACC) were involved in both conditions. Our initial findings have been replicated and extended  extended   [8,21,51,52,58] [8,21,51,52,58].. In a recent functional MRI (fMRI) study, we generalized our previous results on placebo analgesia   [34]   to emotional processing which showed a similar pattern of placebo-dependent activation [36] activation  [36].. Opioid analgesia is mediated through a direct pharmacological act activa ivatio tion n of the en endo dogen genou ouss op opio ioid id recep receptor torss and sha share ress the pr prop op--

  Correspo Corresponding nding author at: MR Research Center N8, Departme Department nt of Clinic Clinical al Neuroscie Neuros cience nce,, Karoli Karolinsk nska a Inst Institu itute, te, 171 76 Stockh Stockholm olm,, Swe Sweden den.. Tel Tel.: .: +46 735101211; fax: +46 8 344146.  [email protected],,  [email protected]  (P. PetroPetroE-mail addresses:  [email protected] vic). *

erties of a placebo manipulation initially. However, differences in the cerebral processing would be expected during ongoing treatment especially in terms of expectation and error signals – processes that have shown to profoundly modulate visual perception [17,28,44,57].. In the opioid condition, and emotional experiences  experiences   [17,28,44,57] expectations for effective analgesia are met and therefore remain congruent with the level of nociceptive processing. On the other ha hand, nd, in the pl place acebo bo co cond nditi ition on,, the treatm treatment ent expect expectati ation on is up uphel held d by means means of the placebo mani manipula pulation tion that is inco incongru ngruent ent with pr proc ocess essin ing g level level of the no nocic cicep eptiv tive e inp input. ut. Po Poten tentia tially lly,, thi thiss wo would uld result in an error signal between the expectations and the nociceptive processin processing. g. The expressio expression n of such an error signal in pain pain processing has been demonstrated by means of deliberate manipulations of the expected intensity of pain [43] pain  [43] where  where the error signa nall co corre rrelat lated ed wi with th an increa increased sed activ activity ity in the later lateral al or orbit bitofr ofron ontal tal cortices. An important question is whether expectations and error processing directly relate to placebo analgesia as has been indi[17,28,44,57]]. cated in other perceptual experiences and emotion  emotion   [17,28,44,57 Behavioral results indicate Behavioral indicate such a relationsh relationship ip as the level of e explixplicit expe expecta ctatio tion n has has bee been n sho shown wn to cor correl relate ate to the subseq subseque uentl ntly y rereported pain level [39] level  [39]..

0304-3959/$36.00     2010 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi: doi:10.1016/j.pain.2010.03.011 10.1016/j.pain.2010.03.011

 

 

60



P. Petro Petrovic vic et al./ PAIN  150 (2010) 59–65

Orbitofr Orb itofronta ontall- and ventrolat ventrolatera erall prefro prefrontal ntal cort cortex ex have been suggested to be involved in higher order emotional and pain regulations [32,35] [32,35]  including including plac placebo ebo ana analgesi lgesia a   [21,27,34,51], [21,27,34,51], and er error ror and expectation expectation processings processings   [18,31,43]. [18,31,43]. However, these regions do not show a clear opioid receptor binding or opioid-induced activation in function functional al imaging imaging studies studies   [10,15,19,34,53,55,56] [10,15,19,34,53,55,56].. In the present pres ent stud study y we hypothe hypothesize sized d that the pre prefro frontal ntal regions regions code code for the expe expected cted nociceptive nociceptive pro processi cessing ng and erro errorr signals, signals, whi which ch then the n infl influen uence ce the descen descendi ding ng op opio ioid id sys system tem in the ACC specifi specifical cally ly during placebo analgesia. In that case, the prefrontal activity and

activity in the PPL condition and not decreased activity in the POP  P -value condition. A threshold of  P  -value (uncor (uncorrect rected) ed) < 0.00 0.001 1 was set for reporting results in rACC and vlPFC/Obfc since our hypothesis involved specifically these regions. We also report when the activity is bordering another region such as anterior insula. In the present re-analysis we also performed a functional connectiv nec tivity ity analys analysis is betw between een the rACC rACC and the vlPFC/ vlPFC/lO lObfc bfc in the plaplacebo (PPL) and the opioid (POP) conditions. In the voxel showing the highest  Z -value -value in the rACC during the main effect of opioids [ X Y Z ] = [8 44 12 12]] the adjuste adjusted d acti activit vity y was ext extra racte cted d from from all concon-

the interaction with the ACC should be more pronounced in the placeb placebo o condit condition ion than than in the op opioi ioid d con condit dition ion.. Mo More reove over, r, the funcfunctional relation between between the prefrontal regions and ACC should correlate with the reported placebo effect. To test this suggestion we re-analyzed the dataset from our previous placebo analgesia PET  [34] and  and emotional placebo fMRI study [36] study  [36].. study [34] study

ditions (Fig. (Fig. 1 1). ). The adjusted activity for each scan was used as a covariate of interest for each condition in a general linear model and the conditions were modeled as confounding covariates. We then the n stu studi died ed the regre regressi ssions ons betwe between en the rACC rACC and the vlPFC/ vlPFC/lO lObfc bfc during the placebo (PPL) and the opioid (POP) conditions. In addition, differences differences in the obse observed rved regres regression sionss between between the cond condiitions were studied by emp employi loying ng a simple simple contrast contrast analysis analysis in SPM99.

2. Methods The data presented here represent a re-analysis of the two previously presented materials  materials   [34,36] [34,36].. The re-analysis involves new contrast and functional connectivity analyses.

 2.1. Study 1 – PET study on placebo and opioid analgesia The first study focused on placebo and opioid analgesia [34] analgesia  [34].. In this positro positron n emissio emission n tom tomogr ography aphy (PET) study study on the placebo placebo analgesia effect nine healthy male subjects participated (age: 20– 27 years). years). Most of the resu results lts have been presen presented ted in a previou previouss  [34].. article [34] article In the present analysis two conditions were further compared: (1) pain and opioid treatment, i.e. opioid analgesia (POP) and (2) pain and placebo placebo treatment, treatment, i.e. placebo placebo analgesia analgesia (PPL (PPL). ). In both conditions the subjects had been stimulated with a tonic noxious heat stimulation (48 C, 70 s, stim stimul ulat atio ion n ar area ea = 2  3 cm) cm) on the the do dors rsumof umof the the le left ft ha hand nd.. In the the PO POP P cond condit itio ion n the the subj subjec ects ts ha had d be been en pre-trea pretreated ted with a shor short-la t-lasting sting opioid opioid (remifen (remifentani tanill 0.5 lg/kg) intra intraven venous ously. ly. In the PP PPL L co cond nditi ition on the sub subjec jects ts had be been en given given sal sal-ine intravenously but they assumed that a strong analgesic was administered. Both treatments were given in a double blind fashion. Both condition conditionss had been per perform formed ed twice twice in each subject subject.. For mor more e detailed detailed method methodss see   [34]. [34]. For completen completeness ess we also modeled the other original conditions in the present re-analysis (Pa (Pain in stimu stimulat lation ion on only ly – P; Warm Warm stimu stimulat latio ion n and op opioi ioid d treatm treatment ent – WOP; Warm stimulation and placebo treatment – WPL; Warm stimulation only – W). However, since these conditions are not a part of the present analysis we will not discuss them further. In the pr prese esent nt re-an re-analy alysis sis we use used d SP SPM9 M99 9 (http://www.fil.ion.ucl.ac.uk/spm)) for pr cl.ac.uk/spm pre-p e-pro roce cessi ssing ng an and d statis statistic tical al ana analys lysis is of the da data. ta. A fixed-effect approach was used for the PET analysis as in the original analysis, i.e. each data point represents a measurement (two for each subject). All conditions (POP, PPL, P, WOP, WPL and W) were entered in the model as in the original analysis. We chose not to re-analyze the dataset in a random effect model since the power was too low due to the limited sample size in the original PE PET T stud study. y. Th The e re resu sult lt ha hass ther theref efor ore e an infe infere renc nce e on only ly on a gr grou oup p le le-vel (i.e. the results are valid for the group of individuals that took pa part rt in the the stud study) y) an and d no nott on a po popu pula lati tion on le leve vel. l. A subt subtra ract ctio ion n an anal al-ysis was performed to test for differences between the PPL and the POP conditions, conditions, i.e. (PPL–PO (PPL–POP) P) and (POP–PP (POP–PPL). L). For the contrast contrast PPL–PO PPL –POP P an exclusive exclusive masking masking pro procedu cedure re was imp implem lemente ented d so that all activations activations induced induced by general general decrease decreasess in the opioid opioid condition vs. the pain condition (i.e. P–POP) with a threshold of   P uncorre unco rrected cted = 0.05 were removed removed.. This procedure procedure indicates indicates that the rem remain aining ing fin findin dings gs in PP PPL– L–PO POP P could could be att attrib ribute uted d to increa increased sed

 2.2. Study 2 – fMRI study on emotional placebo In the second study, the placebo effect was probed during emotional processing by giving a non-specific treatment that subjects believed was an effective anxiolytic substance before presentation of unpleasant pictures [36] pictures  [36].. In this functional magnetic resonance imaging (fMRI) study on emotional placebo effect 15 healthy female mal e subjects subjects par particip ticipated ated (age: 20–3 20–33). 3). On day 1 (expectat (expectation ion inductio indu ction) n) the subj subjects ects watched three bloc blocks ks of unpleasa unpleasant nt and neutral pictures (13 unpleasant and 13 neutral pictures; stimulus duration dura tion was 4 s; inter-stimu inter-stimulus lus interva intervall was 7 s). The expe experien rience ce of unpleasantness was lowered after the subjects received a small intraveno intra venous us dose dose of anxi anxiolyt olytic ic (mi (midaz dazolam olam 0.015 mg/kg) mg/kg) but reversed vers ed after they receiv received ed a bloc blocker ker of the anxiolytic anxiolytic drug (flumazenil maz enil 0.25 mg). mg). This way we induced induced treatme treatment nt exp expecta ectations tions that the anxiolytic drug was highly effective in reducing unpleasantness. On day 2 (placebo testing) the subjects were scanned in a 1.5-T scanner while they watched similar pictures (each block contained 10 unpleasant and 10 neutral pictures; stimulus duration tion wa wass 4 s; inter inter-st -stim imulu uluss inter interval val wa wass jit jitter tered ed bu butt 12.2 12.2 s in averaverage). In some blocks the subjects thought that they had received the anxiolytic drug (but had actually only received saline intravenously) nous ly) and rate rated d the negative pictur pictures es as less unplea unpleasant sant than when they believed that they had received the anxiolytic blocker. Thus, Thu s, in som some e blo blocks cks a pl place acebo bo effect effect had been been accom accompl plish ished ed

ROIss in the rACC rACC use used d for the the functio functional nal conn connect ectivi ivity ty ana analys lysis is are shown shown Fig. 1.   The ROI in the SPM-template below. In Study 1 the ROI was focused on the region showing th the e most most ex expr pres esse sed d acti activa vati tion onss in th the e rACC rACC afte afterr opio opioid id tr trea eatm tmen entt ([ X Y Z ] = [8 44 12]) 12]).. In Stud Study y 2 th the e RO ROII was was fo focu cuse sed d on the most most expre expresse ssed d placebo-ind place bo-induced uced activation activation in the placebo placebo respon responders ders after controlling controlling for unspe unspe-cific placebo effects ([ X Y Z ] = [12 48 12]) 12])..

 



P. Petrovic Petrovic et al./ PAIN  150 (2010) 59–65

(placebo condition) while the other blocks were used as a baseline (contr (control ol con condit dition ion). ). Ta Takin king g the di diffe ffere rent nt visua visuall stimu stimulat lation ionss (unpleas (unp leasant ant and neutral picture pictures) s) into account account,, ther there e were were also four four eve eventnt-rel relate ated d condit conditio ions ns in tot total: al: unp unplea leasan santt pi pictu ctures res and pla pla-cebo (UP), neutral pictures and placebo (NP), unpleasant pictures and control treatment (UC), and neutral pictures and control treatme ment nt (NC) (NC).. Each subject subject went went thro through ugh three sessions sessions,, each one consisting of one placebo block and one control block. For further After er ea each ch blo block ck of pi pictu cture ress the ex expe perie rience nce of  details deta ils see   [36]. [36]. Aft unplea unp leasan santne tness ss was was me measu asured red usi using ng a visual visual ana analo logue gue scale scale (VAS) ranging from 0 (no unpleasantness) to 100 (maximal imagined unpleasantness). The behavioral placebo response was measur sured ed by subtra subtracti cting ng the unp unplea leasa santn ntness ess rat rating ing during during the control condition with the unpleasantness rating during the placebo condition. In the present analysis we used data from the study on emotional placebo described above   [36]   to try (1) to reproduce the findings find ings of a placebo placebo-spe -specific cific func function tional al conn connectiv ectivity ity between between rACC and the vlPFC/lObfc for each subject and (2) to investigate wheth whether er thi thiss fun functi ctiona onall rel relati ation on relat relates es to the de degr gree ee of pl place acebo bo effect (this analysis was not possible for the PET study because the functional connectivity could only be performed on a group level). The activity of the rACC was represented by a region of interest (ROI)) with a radius (ROI radius of 3 mm, mm, centered on the voxel showing showing the highest   Z -score -score in rACC for the placebo responders in the study of emotional placebo ([ X Y Z ] = [12 4812] 4812])) ((Fig. Fig. 1 1;;  [36]  [36]). ). Firstly, we performed within-subject analysis. For each subject, the acti activity, vity, adjusted adjusted for session effects, effects, was extracted extracted for these voxels in the placebo and the control blocks. The average activity for each condition was subtracted from the adjusted activity for each scan (mean correction). These activity values were then used as condition-specific covariates of interest in a GLM implemented in SP SPM9 M99 9 (http://www.fil.ion.ucl.ac.uk/spm ) to gen genera erate te regres regres-sions sions be betwe tween en therACC-R therACC-ROI OI andthe res restt of thebrainduri thebrainduring ng pla placeb cebo o co cond nditi ition on (rPC) (rPC) and con contro troll co condi nditio tion n (rCC), (rCC), in which which the two two treattreatme ment nt conditio conditions ns (placebo (placebo and control control cond condition itions) s) and the four ev event ent-re -relat lated ed condit condition ionss (UP (UP,, NP, NP, UC an and d NC) were were mode modeledas ledas conconfounding foun ding cova covariat riates. es. The func function tional al conn connectiv ectivity ity effe effect ct we were were interested in this analysis was the regression between rACC-ROI and orbitofrontal-/ventrolateral prefrontal cortex in both the placebo condition condition (rPC) (rPC) an and d the control control condition condition (rCC). (rCC). We We also studied the differences between the regressions in placebo condition and control condition (rPC–rCC). The analysis was performed for each subject, however, in two of the subject spikes were detected in one of the three sessions, which therefore had to be removed. Se Secon condly dly,, we pe perfo rform rmed ed a group group ana analys lysis. is. In or orde derr to tes testt whether the functional relation between rACC and prefrontal cortex tex in ea each ch su subj bjec ectt (s (see ee ab abov ove) e) was was re rela late ted d to the the de degr gree ee of pl plac aceb ebo o effec effect, t, a sec second ond-le -level vel rand random om effect effect an analy alysis sis was was pe perfo rform rmed. ed. In thi thiss an analy alysis sis the functi functiona onall connec connectiv tivity ity resul resultt for for ea each ch subjec subjectt was was corcorrelated with the reported placebo effect. This analysis will show whether whe ther the function functional al connectiv connectivity ity betw between een the rAC rACC C and the vlPFC vlPF C is mos mostt pronoun pronounced ced in the best plac placebo ebo responder responders. s. Since we were only interested in the functional relation between rACC and vlPFC/lObfc a threshold of   P -uncorr -uncorrecte ected d < 0.001 was was set for reporti rep orting ng results results in vlPFC vlPFC and lObfc. Since one of the subjects subjects’’ be behav havior ioral al da data ta were were lost lost only only 14 subjec subjects ts were were enr enrol olled led in the sec sec-ond-level ond-lev el analysis.

 

tested whether ACC was more involved in opioid treatment condition, while the prefrontal areas (lObfc and vlPFC) were more involved volv ed in the placebo placebo condition condition.. Moreove Moreover, r, we teste tested d whether whether the functional relation between rACC and the prefrontal regions was more expressed in the placebo condition than in the opioid condition. Analyses assessing the general placebo effect are presented elsewher elsewhere e  [34].  [34].

 3.1.1. Opioid effect vs. placebo effect in pain (POP–PPL; Table 1) The rACC (including the subgenual ACC) was significantly more activated in the opioid condition as compared with the placebo condition. Likewise a region overlapping anterior insula and lObfc was bilaterally bilaterally more activ activated ated in this contrast. contrast. Upon insp inspectio ection n this activation only marginally stretched into lObfc.

 3.1.2. Placebo effect vs. opioid effect in pain (PPL–POP; Table 2) Th The e lO lObfc bfc wa wass signifi significan cantly tly more more act activa ivated ted in the pl place acebo bo condicondition as compared with the opioid condition (Fig. ( Fig. 2 2A). A). This activation was most pronoun pronounced ced in the right hem hemisph isphere ere in term termss of  both the num number ber of activated activated voxels and sign significa ificance nce scores (in  left lObfc this activation did not reach significance). However, the the left no formal comparison was made in the present dataset. The right lObfc activation stretched up into the right vlPFC in a continuum of activated voxels voxels (Fig. 2 2B). B). The placebo-dependent placebo-dependent activations sur surviv vived ed the exclu exclusiv sive e maski masking ng ind indica icatin ting g tha thatt the they y we were re true true act actiivat vatio ions ns in the placeb placebo o co cond nditi ition on rather rather tha than n decre decrease asess in the op opio ioid id condition.  3.1.3. Functional connectivity analysis (Table 3) In this analysis the activity in rACC was co-varied with the rest of the br brain ain in ord order er to cha chara racte cteriz rize e the con connec nectiv tivity ity betw between een rACC rACC rACC C showed showed a signi significa ficant nt co-v co-varia ariation tion and PFC (see Sectio Section n   2). rAC with right and left lObfc and right vlPFC in the PPL condition. No such suc h effect effect wa wass shown shown in the PO POP P con condit ditio ion. n. Wh When en these these two two regressors were compared it was shown that the rACC co-varied significantly more with the right vlPFC and lObfc in the placebo condition than in the opioid condition (Fig. ( Fig. 2 2C C and D).  3.2. Results from Study 2 – fMRI study of emotional placebo A regression analysis was performed on the dataset involving attempt pt to reprod reproduce uce the resu results lts emotional placebo emotional placebo   [36]   in an attem mentioned above indicating a placebo-specific functional relation be betw tween een rACC rACC and vlPFC, vlPFC, and to test test wh wheth ether er thi thiss co conne nnecti ctivit vity y co corrrelated with the placebo response (see Section  2).  2 ). This study has previously shown a placebo-specific increase in rACC and several regions regi ons of lObfc/vl lObfc/vlPFC PFC in the subtracti subtraction on analysis analysis (see further further [36] [36]). ). The placebo response-related activity (UP–UC) in prefrontal cortex cortex also corr correlat elated ed with the beha behavior vioral al plac placebo ebo resp response onse in 3A A and B; the vlPFC vlPFC (but not in the lObfc) lObfc):: [32 46 8]  t   = = 5. 5.89 89 (Fig. (Fig. 3 [36]). [36]).

 Table 1

Regions activated more by opioid (remifentanil 0.5 lg/kg) than by placebo during pain (Study 1). In the present analysis a fixed effect design was used. Region

 3.1. Results from Study 1 – PET study on placebo and opioid analgesia The results results directly directly com compar paring ing the condition condition involving involving pai pain n and opioid opioid trea treatme tment nt (i.e. opioid opioid anal analgesi gesia; a; POP) wi with th pain and plac placebo ebo treatment (i.e. placebo analgesia; PPL) are shown below. Here, we

 

MNI coordinates

 X

3. Results

61

Y

-value t -value

 

-value P -value

Z 

Opioid analgesia vs. placebo analgesia (POP–PPL) rACC (BA 24) Subgenual ACC/mObfc Right anterior insula/lObfc Left anterior insula/lObfc

 

6 0 50 46

52 22 10 4

 

10 14 2 2



4.95 3.48 4.80 5.23

<0.001 <0.001 <0.001 <0.001

(r)ACC, (rostral) (r)ACC, (rostral) anter anterior ior cingul cingulate ate cortex; cortex; mObfc mObfc,, medi medial al orbit orbitofront ofrontal al cortex; cortex; lObfc, lateral orbitofrontal cortex.

 



 

62

P. Petro Petrovic vic et al./ PAIN  150 (2010) 59–65

Fig. 2.  Results from Study 1 in which a fixed effect design was used. The lObfc (A) and the vlPFC (B) were more activated in placebo analgesia than in opioid analgesia (PPL– POP). The The rACC that w was as robustly robustly activated activated also byopioid treatment treatment co-var co-varied ied with the vl vlPFC PFC specifical specifically ly in the placebo placebo condi condition tion as compared compared with tthe he opioid opioid condi condition tion (C). Adjusted rCBF activity indicates this relation for placebo (blue dots) but not for opioid (red dots) treatment (D).

 3.2.1. rACC co-variation (Table 4) Ten of the 15 subjects showed a placebo-specific co-variation Supplem lementa entary ry Tab Table le 1). betwe bet ween en therACC andthe vlPFC/ vlPFC/lOb lObfc fc (see (see Supp In the sec second ond-o -ord rder er rand random om effect effect an analy alysis sis we pr prob obed ed wheth whether er thi thiss functional relation correlated with the behavioral placebo effect. The co-variation between rACC and vlPFC/lObfc tended to be most

binding studies  studies   [19,55,56], [19,55,56], and blood flow studies [1,10,15,53] studies  [1,10,15,53] on  on human subjects indicate that ACC and insula have high concentrations of opioid opioid receptor receptors. s. Altogether, th these ese studies suggest tha thatt the cortical opioid system is involved in mediating placebo analgesia. Indeed, Inde ed, rece recent nt opioid opioid rece receptor ptor binding PET-stud PET-studies ies have show shown n an opioid response in the rACC during placebo analgesia [52,58] analgesia  [52,58]..

expressed for the best placebo responders. No such effects were observed obse rved in the contr control ol condition condition,, and the co-vari co-variatio ation n between between rACC and vlPFC was shown to be significantly stronger in the placebo condition than in the control condition (Fig. (Fig. 3C 3C and D).

Here, we show that opioid-rich areas in rACC and the anterior insula were more extensively activated during opioid as compared with placebo analgesia in line with the stronger analgesic effect in the opioid condition. The activation of these areas was most likely due to direct activation of opioid receptors by the exogenous opioid (remifen (remifentanil tanil). ). Converse Conversely, ly, the resp response onse was opp opposit osite e in the lObfc lOb fc and the vlPFC, vlPFC, tha thatt is, a signifi significan cantly tly larger larger act activa ivatio tion n wa wass ob ob-served for placebo analgesia than opioid analgesia (Fig. (Fig. 2A 2A and B). Raw data suggest that the binding potential in these prefrontal regions is not as high as in ACC and insula  insula   [19,55,56] [19,55,56] and  and they have not sho shown wn increa increased sed act activa ivatio tion n by op opioi ioids ds in other other imagi imaging ng stu studie diess [1,10,15,53] [1,10,15,53].. The previously presented small opioid-induced lObfc [34]   wa wass on only ly ob obse serv rved ed on the the bo bord rder er of an exte extens nsiv ive e ininactivation [34] sula activation. In fact, the activation observed in lObfc may represent sent a sm smoo ooth thin ing g effe effect ct of the the da data ta.. Henc Hence, e, alth althou ough gh it has has no nott been been shown show n whether whether lOb lObfc/vl fc/vlPFC PFC has a low lower er conc concentr entratio ation n of opioid opioid receptors than ACC and insula, the prefrontal regions seem to be

4. Discussion In the present study we extended the analysis of two previous placebo studies in ord placebo order er to characte characterize rize a cognitive cognitive non-op non-opioid ioid component of the placebo response. In the re-analysis of the PET study on placebo placebo analgesi analgesia a [34] we obse observed rved that orbitofr orbitofronta ontall corcortex and vlPFC were were significan significantly tly mo more re activ activated ated during placebo placebo treatme trea tment nt than opio opioid id treatme treatment. nt. Mor Moreove eover, r, rACC disp displaye layed d a funcfunctional connectivity with these prefrontal regions specifically during the placebo context. In the re-analysis of the fMRI study on  we showed that the functional relation beemotional placebo [36] placebo  [36] we tween rACC and vlPFC was associated with the degree of placebo response. In ou ourr or origi iginal nal placeb placebo o analge analgesia sia study study [34] we sho showe wed d a neu neurorophysiological correlate to previous behavioral studies indicating a relation between the endogenous opioid system and placebo analgesia   [2,5–7,26]. gesia [2,5–7,26]. The study showed a co-activated network in opioid and placebo placebo treatments treatments includi including ng the rACC and the anterio anteriorr insula stretching into lObfc. Several other functional imaging studies hav have e also also ind indica icated ted an involv involvem ement ent of rACC rACC in the pl place acebo bo [8,21,51,52,58].. Auto Auto-rad -radiogr iograph aphic ic and post-m post-morte ortem m response   [8,21,51,52,58] studies studi es on hum humans, ans, prima primates tes and rats  [3,38,54],  [3,38,54],   opioid receptor  Table 2

Regions activated more by placebo than by opioid (remifentanil 0.5 pain (Study 1). In the present analysis a fixed effect design was used. Region

 

MNI coordinates

 X

Y

-value t -value

lg/kg)

 

during

-value P -value

Z 

Placebo analgesia vs. opioid analgesia (PPL–POP) Right vlPFC* 36 52 4 * Right lObfc 32 50   12 Left lObfc* 52 24   18

4.43 4.04 3.07

<0.001 <0.001 ns

lObfc, latera laterall orbito orbitofronta frontall corte cortex; x; vlPFC, vlPFC, ventrolateralprefrontal ventrolateralprefrontal corte cortex; x; , indic indicates ates regions surviving the exclusive masking procedure.

involved in an opioid response to a lesser degree and may be more important for other cognitive processes.

 Table 3

Analysis of the functional connectivity between rACC and vlPFC/lObfc in the placebo and opioid conditions (Study 1). In the present analysis a fixed effect design was used. Region Regio n co-varying co-varying with rACC

MNI coord coordinate inatess

[ X Y Z ] = [ 84412 84412]]

 X

Y

Placebo analgesia (PPL) Rig ight ht lObf lObfc/ c/an antterio eriorr insu insula la Right lObfc Right vlPFC Left lObfc/anterior insula Left lObfc  

34 42 36 24 28

18 36 52 14 36

–

–

Opioid analgesia (POP) –

 

 

-value   P -value -value t -value

Z 

 

   

Placebo analgesia vs. opioid analgesia (PPL–POP) Right vlPFC 40 54 Right lObfc 24 18   Left lObfc 42     26

18 4 6 18 14

4.05 3.45 3.90 4.45 4.40

<0.001 <0.001 <0.001 <0.001 <0.001

–

–

–

0 24 20

3.16 3.38 3.60

0.001 <0.001 <0.001

  





rACC, rostral anter anterior ior cingulate cingulate cortex; lObfc, lateral orbito orbitofronta frontall corte cortex; x; vlPFC, vlPFC, ventrolateral prefrontal cortex.

 



P. Petrovic Petrovic et al./ PAIN  150 (2010) 59–65

 

63

Fig. 3.  Results from Study 2 in which a random effect design was used. The placebo-dependent vlPFC activation in (UP–UC) correlated with the subjective placebo response (rating of unple (rating unpleasantn asantness ess after place placebo bo vs. control) control) (A and B). The placebo-sp placebo-specific ecific rACC co-va co-variatio riation n with the vlPFC correl correlated ated also with the subjective subjective place placebo bo rating ratingss (C and D).

Functional neuroimaging neuroimaging and neurophysiological neurophysiological studies suggest that the Obfc stores representations of the motivational value for specific primary primary and secondary secondary re-enforcers re-enforcers   [18,22,23,30,31, 45,46].. The monitor 45,46] monitoring ing function function involves involves represen representatio tation n of reward rewardss [30,31,41,45,46].. Contemand expectations of upcoming rewards  rewards   [30,31,41,45,46] porary models of Obfc function suggest that one key function of  this region is to signal the desirability of the expected outcomes and hold those asso associati ciations ons onon-line line until the mo motivati tivational onal goal has bee been n achi achieved eved [42] [42].. Mor Moreove eover, r, pre prefron frontal tal regions regions are imp importa ortant nt in gene generati rating ng predicti prediction on err errors ors in associati associative ve learning learning [47] an and d also also in pain processing [43] processing  [43].. Similar cognitive processes are involved in the placebo mechanism [14] [14].. Ex Expl plici icitt ex expe pecta ctatio tions ns of a re relat lative ive rew rewar ard d (i. (i.e. e. the ex expe pecctation tation of pa pain in rel relief ief)) is a ke key y fun functi ction on in the pl place acebo bo analge analgesic sic response   [29,49,50], response [29,49,50], and treatment expectations correlate strongly with the subsequent placebo response  [39]  [39].. We therefore suggest that the involvem involvement ent of these pre prefron frontal tal regions regions in plac placebo ebo anal analgegesia is rel relate ated d to the ex expe pecta ctatio tion n of a treatm treatment ent effect effect.. A po possi ssible ble cricritique of this suggestion is that also the opioid analgesia condition should contain treatment expectatio expectation, n, and and therefore therefore activate these prefrontal regions to a similar degree. This is not challengeable in the anticipat anticipation ion phase. phase. How However ever,, we propose propose that neith neither er the expectation nor the error signals are the same during the two con-

 Table 4

Study of the function functional al conn connectivi ectivity ty between between rACC and right vlPFC/lObfc vlPFC/lObfc that correlated with the subjective placebo response (Study 2). In the present analysis a random effect design was used. Region Regio n co-var co-varying ying with rACC

MNI coordinate coordinatess

[ X Y Z ] = [1248 12] 12]

 X

Y

Placebo condition (rPC) Right vlPFC/lObfc Right vlPFC/lObfc Left lObfc/lObfc

28 34 26

54 54 54 54

–

–

 

Control condition (rCC) –

 

-value   P -value -value t -value

Z  4 8 6

3.36 2.73 3.11

–

–

–

 



 



4.36 3.25 3.24

<0.001 0.003 ns 0.004 ns

 

 



0.003 ns 0.009 ns 0.005 ns

Placebo condition vs. control condition (rPC–rCC) Right vlPFC/lObfc Left vlPFC/lObfc Left lObfc/lObfc

   

28 26 28



52 52 50 50

 

4 6 6

rACC, rostral anteri anterior or cingulate cingulate corte cortex; x; lObfc, lateral orbito orbitofronta frontall corte cortex; x; vlPFC vlPFC,, ventrolate ventr olateral ral prefro prefrontal ntal corte cortex; x; rCC, regres regression sion analysis analysis during control condition; condition; rPC, regression analysis during placebo condition.

ditions, since the opioid effect reduces the nociceptive input directl rectly y thr throu ough gh op opioi ioid d recep receptor tor act activa ivatio tion n wh while ile the placeb placebo o condition does not. In other words the subjects still expect a pain reduction during this condition, and will have a stronger error signa nall since since the expect expectati ations ons do not match match wi with th the no nocic cicep eptiv tive e inp input. ut. This suggests that the vlPFC/Obfc may be more active in the placebo condition (due to stronger expectations and larger error signal) in line with our findings. An important question is whether this prefrontal activity processing expectations and the error signal may also be involved in inducing indu cing the plac placebo ebo response. response. It has been show shown n that expectaexpectations tions hav have e a majo majorr impa impact ct on visual visual perce perceptu ptual al exp experi erienc ence e [17,44,57] on affe affective ctive exp experie erience nce [28], [28], and elabo elabora ratin ting g up upon on those those expectations requires the function of an intact Obfc [9,12] Obfc  [9,12].. Moreover, vlPFC and the lObfc have been implicated in top-down modulation of pain [4,33,40] pain  [4,33,40] including  including placebo analgesia  analgesia   [21,27,34,51], [21,27,34,51], depression   [13]   and exp experim erimenta entall anxi anxiety ety   [20,24,25,32]. [20,24,25,32]. Th These ese studies are in line with the idea that lateral ventro-orbital cortex is involved in a general, and not only pain specific, cognitive pro [35].. cess that mediates modulation of pain and emotion  [35] A model for the mechanisms driving the placebo effect using expectations and error processing is that this is a consequence of  Bayesian processing – as has been suggested for other perceptual [16,17,57].. In that hyp hypothe othesis sis percept perceptions ions derive derived d from from systems   [16,17,57] external signals are a compromise between incoming signals and the expe expecta ctatio tions ns in the sys system tem.. It has bee been n sug sugges gested ted tha thatt thi thiss type type of process processing ing vastl vastly y increase increasess effic efficienc iency y of understa understandin nding g the outer outer world. While lObfc and vlPFC may be involved in cognitive processes dri driving ving the anal analgesi gesic c response response dur during ing plac placebo, ebo, several several rece receptor ptor and functional imaging studies have not found the same type of  an op opio ioidid-rel relate ated d invol involvem vement ent as shown shown for for AC ACC C and ins insula ula [10,15,19,34,53,55,56] . Here, Here, we suggest that prefron prefrontal tal regi regions ons ma may y influence influence opioi opioid d syste systems ms in rAC rACC C – possibly possibly throug through h local local enkephalinergic inter-neurons. It has been suggested that such inter-neurons may activate pyramidal projection neurons (through opioid receptors) that in turn may regulate distant pain-modulating areas are   [48]. [48] . In co-va line with our hypo hypothes is, our regre ssion n data show showed edasthe rACC co-varied ried with the thesis, lOb lObfc/vl fc/vlPFC PFCregressio in the placebo analgesia condition but not in the opioid analgesia condition. The same sam e reg region ionss that that coco-var varied ied with the rACC rACC we were re also also a part part of  the regions that were specifically activated during placebo treatment (PPL–POP) (PPL–POP) (Fig. 2 2B). B).   We used the data from the study on

 

64

 



P. Petro Petrovic vic et al./ PAIN  150 (2010) 59–65

emotional placebo emotional placebo   [36]  [36]   to probe whether this functional relation was directly related to the subjectively reported placebo response. In that that st stud udy y we ha have ve pr prev evio ious usly ly sh show own n that that bo both th the the rA rACC CC an and d the the vlPFC correlate with the subjectively reported placebo effect and  [36].. In the present analysis we show that treatment expectations expectations [36] the subjects who expressed the highest placebo response also had the most reliable placebo-specific coupling between the rACC and the the vlPF vlPFC C (Fig. Fig. 3C and D). D). Th Thus, us, we bo both th replic replicate ated d the pl place acebo bo-s -spepecific functional relation between rACC and vlPFC and showed that it is dir direct ectly ly rel relate ated d to the pl place acebo bo effect effect.. The fun functi ctiona onall conne connecti ctivvity anal analysis ysis in the emo emotion tional al plac placebo ebo study gene generali ralizes zes the idea that prefrontal processes may interact with neuromodulatory systemss in ACC to oth tem other er ar area eass than than pa pain. in. Fur Furthe therm rmor ore, e, the the fin findin ding g tha thatt activity in rACC may be opioid dependent also in emotional processing [37] cessing  [37] suggests  suggests that the prefrontal cortex may drive opioidrelated modulatory systems in ACC during emotional placebo as well. There are several alternative interpretations of the present results. Firstly, it cannot be excluded that a placebo-like increase in lObfc lOb fc was was ma maske sked d in the op opio ioid id con condit dition ion by ph phar arma macol colog ogica ically lly induced decreases in CBF that is either neural or vascular in origin. Second Sec ondly, ly, connec connectiv tivity ity ana analys lysis is do does es no nott show show any cau causal sality ity – rACC rACC may as well control prefrontal activity. It is also possible that the functional relation between rACC and lObfc represents a cognitive bias to rep report ort pa pain in scor scores es in line with with one’s one’s prio priorr beli beliefs. efs. Moreo Moreover, ver, the lack lack of rACC– rACC–Ob Obfc fc cor corre relat lationin ionin theopioid theopioid condit conditionmay ionmay be ex ex-plained by other factors than a fulfilled expectation, e.g. it would also also be po possi ssible ble tha thatt a red reduce uced d level level of consc consciou iousne sness ss or the he heigh ighttened euphoric level during opioid condition attenuates this functional relation. Finally, we did not measure expectations although we use used d the rep repor orted ted de decre crease ase in unp unplea leasan santne tness ss during during the con condidi[36].. tioning day (day 1) as a proxy to treatment expectations in   [36] Interestingly, this value correlated to the placebo effect day 2 in both ACC and vlPFC. Finally, we only used open treatment in the op opio ioid id condit condition ion an and d it would would be inter interest esting ing to study study the pr prefr efron ontal tal involvement in open opioid treatment vs. hidden opioid treatment [11] [11] associated  associated with different degrees of treatment expectation. In conclu conclusio sion, n, the pr prese esent nt res result ultss show show tha thatt pl place acebo bo an analg algesi esia a is qualitatively different from an opioid drug response during pain. These The se fin findin dings gs and the sugge suggesti stion on that that dr drugs ugs dir direct ectly ly inter interfer fere e with with the exp expectat ectation ion path pathway way [11] ind indica icate te tha thatt a pl place acebo bo resp respons onse e ma may y not be interpreted as a passive control to a specific drug effect, but a highly active state in itself.

Conflict of interest The authors of this manuscript have no financial or other relationship that may be perceived as a conflict of interest.

 Acknowledgments The research was supported by grants from the EFIC-Grünenthall Grant tha Grant,, Sw Swed edish ish Me Medi dical cal Re Resea search rch Co Counc uncil, il, the Kar Karol olins inska ka InstiInstitute tute,, BB BB/C /CBM BME, E, LA LA,, FE FEDE DER/ R/PO POCI CI 20 2010 10,, th the e Kn Knut ut an and d Al Alic ice e Wallenb Wal lenberg erg Foun Foundati dation, on, tthe he Barb Barbro ro and Bern Bernard ard Oshe Osherr Foundati Foundation, on, The Swedish Governmental Agency for Innovation Systems (VINNOVA NOVA), ), Th The e Sw Swed edish ish Fo Found undati ation on for for Str Strate ategic gic Re Resea search rch and the Pe Pettrus and Augusta Hedlunds Foundation. The authors are grateful to all participants from the MRC/PET facility and for all suggestions and comments from colleagues.

 Appendix A. Supplementary data Suppl Sup plem ement entary ary da data ta ass associ ociate ated d with with thi thiss ar artic ticle le can can be fou found nd,, in the online version, at  at   doi:10.1016/j.pain.2010.03.011. doi:10.1016/j.pain.2010.03.011.

References [1] Adler LJ, Gyulai FE, Diehl DJ, Mint Mintun un MA, Wint Winter er PM, Fires Firestone tone LL. Regional Regional brain brain activi activity ty cha change ngess asso associa ciated ted wit with h fen fentan tanyl yl ana analge lgesia sia elu elucid cidate ated d by positron emission tomography. Anesth Analg 1997;84:120–6. [2] Ama Amanzi nzio o M, Ben Benede edetti tti F. Neurop Neurophar harma macol cologi ogical cal dissec dissectio tion n of placeb placebo o ana analge lgesia sia:: expect expectati ationon-act activa ivated ted opioid opioid sys system temss versus versus con condit dition ioning ing-activated specific subsystems. J Neurosci 1999;19:484–94. [3] Atweh SF, Kuhar MJ. Distri Distributio bution n and physiological physiological significa significance nce of opioi opioid d receptors in the brain. Br Med Bull 1983;39:47–52. [4] Bantick SJ, Wise RG, Ploghaus A, Clare S, Smith SM, Tracey Tracey I. Imagi Imaging ng how attention modulates pain in humans using functional MRI. Brain 2002;125: 310–9. [5] Bened Benedetti etti F. The opposite opposite effects of the opiate antagonist antagonist naloxon naloxone e and the cholecysto chole cystokinin kinin antagonist proglumid proglumide e on place placebo bo analg analgesia. esia. Pain 1996;6 1996;64: 4: 535–43. [6] Benede Benedetti tti F, Amanzi Amanzio o M, Baldi Baldi S, Cas Casadi adio o C, Maggi Maggi G. Induci Inducing ng placeb placebo o respir respirato atory ry depres depressan santt respon responses ses in humans humans via opioid opioid recept receptors ors.. Eur J Neurosci 1999;11:625–31. [7] Benedetti F, Arduino C, Amanzio M. Somatotopic activation of opioid systems by target-directed expectations of analgesia. J Neurosci 1999;19:3639–48. [8] Bingel U, Lorenz J, Gläsc Gläscher her J, Weiller Weiller C, Büchel Büchel C. Mechanism Mechanismss of placebo placebo analgesia: analg esia: rACC recruitme recruitment nt of a subcortical subcortical antinocic antinociceptiv eptive e netwo network. rk. Pain 2006;120:8–15. [9] Camille Camille N, Corice Coricelli lli G, Sal Sallet let J, Pradat Pradat-Di -Diehl ehl P, Duh Duhame amell J-R, J-R, Sirigu Sirigu A. The involv involveme ement nt of orbito orbitofro fronta ntall cor cortex tex in the experi experienc ence e of reg regret ret.. Scienc Science e 2004;304:1167–70. [10] Casey KL, Svensson Svensson P, Morrow Morrow TJ, Raz J, Jone C, Minoshi Minoshima ma S. Selective opiat opiate e modulation modu lation of nociceptive nociceptive process processing ing in the human brain. brain. J Neuro Neurophysi physiol ol 2000;84:525–33. [11] Colloca Colloca L, Bened Benedetti etti F. Placebos and painki painkillers: llers: is mind as real as matter? Nat Rev Neurosci 2005;6:545–52. [12] Coricelli Coricelli G, Critchley HD, Joffily M, O’Dohe O’Doherty rty J, Sirigu A, Dolan RJ. Regret and its avoida avoidance nce:: a neu neuroi roima magin ging g stu study dy of cho choice ice beh behavi avior. or. Nat Neu Neurosc roscii 2005;8:1255–62. [13] Drevets Drevets WC. Neuroimagi Neuroimaging ng and neuropathol neuropathologica ogicall studi studies es of depre depression: ssion: implicati impl ications ons for the cognitivecognitive-emoti emotional onal featur features es of mooddisorders. Curr Opin Neurobiol 2001;11:240–9. [1 [14] 4] Fie Field ldss H. St Stat atee-de depe pend nden entt opio opioid id co cont ntro roll of pain pain.. Na Natt Re Rev v Ne Neur urosc oscii 2005;5:565–75. [15] Firestone Firestone LL, Gyulai F, Mintun M, Adler LJ, Urso K, Winter PM. Human brain activity activ ity response response to fentanyl fentanyl imaged imaged by posit positron ron emiss emission ion tomography tomography.. Anest Anesth h Analg 1996;82:1247–51. [16] Friston K. A theor theory y of cortical cortical responses. responses. Philos Trans R Soc Lond B Biol Sci 2005;360:815–36. [1 [17] 7] Frit Frith h CD CD.. Maki Making ng up th the e mind mind:: ho how w th the e brai brain n crea create tess ou ourr ment mental al world. Blackwell Publishing; 2007. [18] [18] Got Gottfri tfried ed JA, O’D O’Dohe ohertyJ, rtyJ, Dol Dolan an RJ. Encodingpredi Encodingpredicti ctive ve rewardvalue rewardvalue in hum human an amygdala and orbitofrontal cortex. Science 2003;301:1104–7. [19] Jones AK, Qi LY, Fujirawa T, Luthra SK, SK, Ashburner J, Bloomfield P, Cunningham VJ, Itoh M, Fukuda H, Jones T. In vivo distribution of opioid receptors in man in relation to the cortical projections of the medial and lateral pain systems measured meas ured with positron positron emiss emission ion tomography tomography.. Neuro Neurosci sci Lett 1991;126: 25–8. [20] Kalisch R, Wiech K, Critchley HD, Seymour Seymour B, O’Doherty JP, Oakley DA, Allen P, Dolan RJ. Anxiety Anxiety reduction reduction through through detachment detachment:: subjective, subjective, physi physiologi ological, cal, and neural effects. effects. J Cogn Neurosci 2005;17:8 2005;17:874–83. 74–83. [21] [21] Kon Kong g J, Gollub Gollub RL,Rosma RL,Rosman n IS, WebbJM, WebbJM, Van Vangel gel MG, MG, Kirs Kirsch ch I, Ka Kaptc ptchukTJ. hukTJ. Brain activity associated with expectancy-enhanced placebo analgesia as measured by functional magnetic resonance imaging. J Neurosci 2006;26:381–8. [22] Kringelbach Kringelbach ML, Rolls ET. Neura Neurall correlates of rapid rever reversal sal learning in a simple model of human social interaction. NeuroImage 2003;20:1371–83. [23] [23] Kri Kringe ngelba lbach ch ML, ML, Rolls Rolls ET. The fun functi ctiona onall neu neuroa roanat natomy omy of the human human orbitofront orbit ofrontal al cortex: cortex: evidence evidence from neuroi neuroimagin maging g and neurop neuropsycho sychology. logy. Prog Neurobiol 2004;72:341–72. [24] Levesque Levesque J, Eugen Eugene e F, Joanette Joanette Y, Paquette V, Mensour B, Beaudoin G, Leroux  JM, Bourg Bourgouin ouin P, Beauregard Beauregard M. Neural circui circuitry try under underlying lying voluntary voluntary suppression of sadness. Biol Psychiatry 2003;53:502–10. [25] [25] Levesq Levesque ue J, Joanet Joanette te Y, Menso Mensour ur B, Bea Beaudo udoin in G, Ler Leroux oux JM, Bou Bourgo rgouin uin P, Bea Beaure uregar gard d M. Neu Neural ral bas basis is of emo emotio tional nal self-re self-regul gulati ation on in chi childh ldhood ood.. Neuroscience 2004;129:361–9. [26] Levine Levine JD, Gord Gordon on NC, Fields Fields HL. The mechanism mechanism of placebo placebo analgesia analgesia.. Lance Lancett 1978;2:654–7. [27] [27] Lieber LiebermanMD, manMD, JarchoJM, JarchoJM, Ber BermanS, manS, Nalibo Naliboff ff BD, Suy Suyeno enobu bu BY,Mande BY,Mandelke lkern rn M, Mayer May er EA. The neural correlate correlatess of placebo effects: a disrup disruption tion account. Neuroimage 2004;22:447–55. [28] Mellers BA. Choic Choice e and the relative relative pleasure of conse consequence quences. s. Psych Psychol ol Bull 2000;126:910–24. [29] Montgome Montgomery ry GH, Kirsch I. Classi Classical cal conditioning conditioning and the placebo placebo effect. Pain 1997;72:107–13. [30] O’Doherty O’Doherty J, Kring Kringelbac elbach h ML, Rolls ET, Hornak J, Andrews Andrews C. Abstr Abstract act reward and punishment punishment representatio representations ns in the human orbitofront orbitofrontal al cortex. cortex. Nat Neurosci 2001;4:95–102. [31] O’Doherty O’Doherty JP, Deichmann R, Critchley HD, Dolan RJ. Neural Neural responses during anticipation of a primary taste reward. Neuron 2002;33:815–26.

 



P. Petrovic Petrovic et al./ PAIN  150 (2010) 59–65

 

65

[32] Ochsner KN, Bunge SA, Gross JJ, Gabrieli Gabrieli JD. Rethi Rethinking nking feeling feelings: s: an FMR FMRII study study of the cog cognit nitive ive regula regulatio tion n of emo emotio tion. n. J Cog Cogn n Neuros Neurosci ci 2002;1 2002;14: 4: 1215–29. [33] Petrovic Petrovic P, Peters Petersson son KM, Ghatan PH, Stone-Elan Stone-Elander der S, Ingvar M. Pain-relat Pain-related ed cerebral cerebr al activatio activation n is altered by a distr distractin acting g cognitive cognitive task. Pain 2000 2000;85: ;85: 19–30. [34] Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and opioid analgesia analgesia – imagine a shared neuronal network. Science 2002;295:1737–40. [35] Petrovic Petrovic P. Opioi Opioid d and place placebo bo analgesia share the same network. network. Semin Pain Med 2005;3:31–6. [36] Petrovic Petrovic P, Dietr Dietrich ich T, Frans Fransson son P, Ander Andersson sson J, Carlsson Carlsson K, Ingva Ingvarr M. Place Placebo bo in emotio emo tional nal pro proces cessin sing-i g-indu nduced ced exp expect ectati ations ons of anx anxiet iety y relief relief act activa ivate te a generalized modulatory network. Neuron 2005;46:957–69.

[46] Tremblay Tremblay L, Schultz Schultz W. Relative Relative reward prefe preference rence in prima primate te orbitofront orbitofrontal al cortex [see comments]. Nature 1999;398:704–8. [47] Turner DC, Aitken MRF, Shanks DR, Sahakian BJ, Robbins TW, Schwarzbauer C, Fletcher PC. The role of lateral frontal cortex in causal associative learning: exploring preventative and super-learning. Cereb Cortex 2004;14:872–80. [48] [48] Vog Vogtt BA, Fin Finch ch DM DM,, Ols Olson on CR. Function Functional al het hetero erogen geneit eity y in cingul cingulatecorte atecortex: x: the anterior anteri or execu executive tive and poste posterior rior evaluativ evaluative e region regions. s. Cereb Cortex 1992;2: 435–43. [49] Voudouris Voudouris NJ, Peck CL, Coleman G. Conditione Conditioned d respo response nse models of place placebo bo phenomena: further support. Pain 1989;38:109–16. [50] [50] Voud Voudour ouris is NJ, Peck Peck CL, Colema Coleman n G. The role role of con condit dition ioning ing and verbal verbal expectancy in the placebo response. Pain 1990;43:121–8. [51] Wager Wager TD, Rilling Rilling JK, Smith EE, Sokol Sokolik ik A, Casey KL, David Davidson son RJ, Kosslyn SM,

[37] Petrovic P, Pleger B, Seymour B, Kloppel S, De Martino B, Critchley H, Dolan RJ. Blocking Block ing central central opiate opiate function function modu modulates lates hedonic impact impact and anterior cingulate response to rewards and losses. J Neurosci 2008;28:10509–16. [38] Pfeiffer Pfeiffer A, Pasi A, Mehraein P, Herz A. Opiate Opiate receptor binding binding sites in human brain. Brain Res 1982;248:87–96. [39] Price DD, Milling Milling LS, Kirsch I, Duf FA, Montgom Montgomery ery GH, Nichol Nicholls ls SS. An analysis of factor factorss that that con contri tribut bute e to the magni magnitud tude e of placeb placebo o ana analge lgesia sia in an experimental paradigm. Pain 1999;83:147–56. [40] Rainville Rainville P, Hofbauer RK, Paus T, Duncan Duncan GH, Bushne Bushnell ll MC, Price DD. Cerebral Cerebral mechanism mech anismss of hypnotic hypnotic induction and sugge suggestion. stion. J Cogn Neurosci 1999;11: 110–25. [41] Schoenbaum Schoenbaum G, Chiba AA, Galla Gallagher gher M. Orbit Orbitofront ofrontal al cortex cortex and basolateral basolateral amygdala amyg dala encode expected expected outcomes outcomes during learning. learning. Nat Neuro Neurosci sci 1998;1 1998;1:: 155–9. [42] Schoenbaum Schoenbaum G, Roesc Roesch h M. Orbito Orbitofronta frontall cortex, cortex, associative associative learning, and expectancies. Neuron 2005;47:633–6. [43] Seymour Seymour B, O’Doherty O’Doherty JP, Kolt Koltzenbur zenburg g M, Wiech K, Frackowiak Frackowiak R, Friston K, Dolan R. Oppo Opponent nent appetitiveappetitive-avers aversive ive neural processes underlie underlie predictive predictive learning of pain relief. Nat Neurosci 2005;8:1234–40. [44] Sterzer P, Frith C, Petro Petrovic vic P. Believ Believing ing is seeing: seeing: expe expectati ctations ons alter visual awareness. Curr Biol 2008;18:R697–8. [45] Thorpe SJ, Rolls ET, Maddison Maddison S. The orbitofron orbitofrontal tal cortex: neurona neuronall activity in

Rose RM, Cohen JD. Placebo-induced changes in FMRI in the anticipation and experience of pain. Science 2004;303:1162–7. [52] Wager Wager TD, Scott DJ, Zubieta Zubieta JK. Placebo effects on human mu-op mu-opioid ioid activity activity during pain. Proc Natl Acad Sci USA 2007;104:11056–61. [53] [53] Wagne Wagnerr KJ, Willoc Willoch h F, Kochs Kochs EF, Siessm Siessmeie eierr T, Tölle Tölle TR. Dos Dose-d e-depe epende ndent nt regional cerebral blood flow changes during remifentanil infusion in humans. A positron emission tomography study. Anesthesiology 2001;94:732–9. [54] [54] Wamsle Wamsley y JK, Zar Zarbin bin MA, MA, You Young ng 3rd WS, Kuh Kuhar ar MJ. Dis Distri tribut bution ion of opi opiate ate receptors recep tors in the monkey monkey brain: brain: an autoradiog autoradiographi raphic c study study.. Neuroscienc Neuroscience e 1982;7:595–613. [55] Willoch F, Tolle TR, Wester HJ, Munz F, Petzold A, Schwaiger M, Conrad B, Bartenstein P. Central pain after pontine infarction is associated with changes in opioid receptor binding: a PET study with   11C-diprenorphine. AJNR Am J Neuroradiol 1999;20:686–90. [56] Willoch Willoch F, Schindler F, Wester HJ, Empl M, Straube A, Schwaiger M, Conrad B, Tolle TR. Central poststroke pain and reduced opioid receptor binding within pain processing circuitries: a [11C] dipre diprenorphi norphine ne PET study. Pain 2004;108: 2004;108: 213–20. [57] [57] Yui YuilleA, lleA, Kerste Kersten n D. Vision Vision as Bay Bayesi esian an inferen inference:analy ce:analysis sis by syn synthe thesis?Trend sis?Trendss Cogn Sci 2006;10:301–8. [58] [58] Zubi ZubietaJK, etaJK, Bue Buelle llerr JA, Jackso Jackson n LR,Scott DJ, Xu Y, KoeppeRA, KoeppeRA, Nic Nichol holss TE, Stohle Stohlerr CS. Placebo Placebo effect effectss medi mediated ated by endogenous endogenous opioid activity on mu-op mu-opioid ioid

the behaving monkey. Exp Brain Res 1983;49:93–115.

receptors. J Neurosci 2005;25:7754–62.