12014 • The Journal of Neuroscience, November 15, 2006 • 26(46):12014 –12022
`
`Behavior/Systems/Cognitive
`
`The Biochemical and Neuroendocrine Bases of the
`Hyperalgesic Nocebo Effect
`
`Fabrizio Benedetti,1 Martina Amanzio,2 Sergio Vighetti,1 and Giovanni Asteggiano3
`1Department of Neuroscience, University of Turin Medical School, 10125 Turin, Italy, 2Department of Psychology, University of Turin, 10124 Turin, Italy,
`and 3Division of Neurology, S. Lazzaro Medical Center, 12051 Alba, Italy
`
`Despite the increasing research on placebos in recent times, little is known about the nocebo effect, a phenomenon that is opposite to the
`placebo effect and whereby expectations of symptom worsening play a crucial role. By studying experimental ischemic arm pain in
`healthy volunteers and by using a neuropharmacological approach, we found that verbally induced nocebo hyperalgesia was associated
`to hyperactivity of the hypothalamic–pituitary–adrenal (HPA) axis, as assessed by means of adrenocorticotropic hormone and cortisol
`plasma concentrations. Both nocebo hyperalgesia and HPA hyperactivity were antagonized by the benzodiazepine diazepam, suggesting
`that anxiety played a major role in these effects. The administration of the mixed cholecystokinin (CCK) type-A/B receptor antagonist
`proglumide blocked nocebo hyperalgesia completely but had no effect on HPA hyperactivity, which suggests a specific involvement of
`CCK in the hyperalgesic but not in the anxiety component of the nocebo effect. Importantly, both diazepam and proglumide did not show
`analgesic properties on basal pain, because they acted only on the nocebo-induced pain increase. These data indicate a close relationship
`between anxiety and nocebo hyperalgesia, in which the CCKergic systems play a key role in anxiety-induced hyperalgesia. These results,
`together with previous findings showing that placebo analgesia is mediated by endogenous opioids, suggest that the analgesic placebo/
`hyperalgesic nocebo phenomenon may involve the opposite activation of endogenous opioidergic and CCKergic systems.
`
`Key words: nocebo; placebo; pain; hypothalamus–pituitary–adrenal axis; benzodiazepines; cholecystokinin
`
`Introduction
`In recent times, the placebo effect has been analyzed with sophis-
`ticated neurobiological tools that have uncovered specific mech-
`anisms at the biochemical, cellular, and anatomical level in dif-
`ferent systems and conditions, such as pain, motor disorders,
`depression, and immune– endocrine responses (Benedetti et al.,
`2005; Colloca and Benedetti, 2005). It has been shown that this
`may occur through both expectation and conditioning mecha-
`nisms, although expectations and emotions seem to play a fun-
`damental role (Benedetti et al., 2003; Price et al., 2005). Most of
`our knowledge about the placebo effect comes from the field of
`pain, in which both a neuropharmacological approach with opi-
`oid antagonists (Levine et al., 1978; Grevert et al., 1983; Levine
`and Gordon, 1984; Amanzio and Benedetti, 1999; Benedetti et al.,
`1999; Hoffman et al., 2005) and, more recently, brain imaging
`techniques (Petrovic et al., 2002, 2005; Wager et al., 2004, 2006;
`Zubieta et al., 2005; Keltner et al., 2006; Kong et al., 2006) have
`been used.
`In contrast, the neurobiological mechanisms of the nocebo
`effect have been less investigated, despite them being interesting
`
`Received July 12, 2006; revised Sept. 10, 2006; accepted Oct. 11, 2006.
`This work was supported by Project “Neuroscience” of the National Research Council Grants 01.00439.ST97 and
`02.00529.ST97, Project “Alzheimer’s disease” of the Italian Ministry of Health Grants PFA/DML/UO6/2001 and PFA/
`DML/UO6/2001/AA, and Italian Ministry of University and Research–Fondo per gli Investimenti della Ricerca di Base
`Grant RBNE01SZB.
`Correspondence should be addressed to Fabrizio Benedetti, Dipartimento di Neuroscienze, Universita` di Torino,
`Corso Raffaello 30, 10125 Torino, Italy. E-mail:fabrizio.benedetti@unito.it
`DOI:10.1523/JNEUROSCI.2947-06.2006
`Copyright © 2006 Society for Neuroscience
`
`0270-6474/06/2612014-09$15.00/0
`
`as those of the placebo effect. For example, hyperalgesia after
`expectation of painful stimulation is associated with changes in
`brain activation of different regions (Sawamoto et al., 2000;
`Koyama et al., 2005; Keltner et al., 2006). Expectation and/or
`conditioning mechanisms, which are similar and opposite to
`those of the placebo counterpart, are supposed to be involved
`(Benedetti et al., 2003). In a previous clinical study, we showed
`that nocebo hyperalgesia could be prevented by pretreatment
`with proglumide, a nonspecific cholecystokinin (CCK) antago-
`nist for both CCK-A and CCK-B receptors, suggesting the possi-
`ble involvement of CCKergic systems in the nocebo effect
`(Benedetti et al., 1997). However, because of ethical constraints
`in these patients, we did not have the possibility to investigate
`these effects further.
`By using selective CCK-A and CCK-B receptor antagonists, sev-
`eral studies in animals and humans have shown the important role
`of CCKergic systems in the modulation of anxiety and in the link
`between anxiety and hyperalgesia (Hebb et al., 2005). For example,
`in a social-defeat model of anxiety in rats, it has been shown
`recently that CI-988 (4-[[2-[[3-(1H-indol-3-yl)-2-methyl-1-oxo-2
`[[(tricyclo[3.3[12,17]dec-2-yloxy)-carbonyl]amino]-propyl]amino]-1-
`phenyethyl]amino]-4-oxo-[R-(R*, R*)]-butanoate N-methyl-D-
`glucamine), a selective CCK-B receptor antagonist, prevented
`anxiety-induced hyperalgesia, with an effect that was similar to that
`produced by the established anxiolytic chlordiazepoxide (Andre et
`al., 2005).
`By taking all of these considerations into account and by con-
`sidering the ethical limitations in the patients of our previous
`
`SPRUCE - EXHIBIT 2005
`Neurocrine Biosciences, Inc. v. Spruce Biosciences, Inc. - PGR2022-00025
`
`

`

`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`J. Neurosci., November 15, 2006 • 26(46):12014 –12022 • 12015
`
`Table 1. Characteristics (mean ⴞ SD) of the subjects of each experimental group
`Sex (M/F)
`Age (years)
`Weight (kg)
`37.4 ⫾ 12.4
`63.1 ⫾ 11.5
`36.9 ⫾ 13.8
`62.7 ⫾ 13.5
`39.1 ⫾ 13.3
`64.0 ⫾ 13.1
`37.1 ⫾ 11.9
`66.1 ⫾ 14.6
`
`Group 1
`Group 2
`Group 3
`Group 4
`
`6/6
`6/7
`6/6
`5/7
`
`Basal ACTH (pg/ml)
`22.7 ⫾ 3.4
`20.5 ⫾ 4.4
`19.4 ⫾ 3.4
`20.4 ⫾ 3.7
`
`Basal cortisol (␮g/L)
`106.4 ⫾ 8.7
`106.5 ⫾ 8.9
`103.8 ⫾ 7.8
`104.8 ⫾ 8.3
`
`STAI-S
`40.6 ⫾ 4.3
`41.5 ⫾ 5.2
`41.6 ⫾ 6.2
`40.7 ⫾ 5.5
`
`STAI-T
`41.1 ⫾ 5.9
`40.1 ⫾ 4.7
`39.2 ⫾ 4.7
`40.6 ⫾ 5.3
`
`M, Male; F, female.
`
`study (Benedetti et al., 1997), we performed a detailed neuro-
`pharmacological analysis of the hyperalgesic nocebo effect in
`healthy volunteers by using the nonselective CCK-A and CCK-B
`receptor antagonist proglumide. We decided to start this phar-
`macological investigation of nocebo hyperalgesia because the
`neuropharmacological approach in humans has represented a
`crucial step to unravel the opioid mechanisms of placebo analge-
`sia (Levine et al., 1978; Grevert et al., 1983; Levine and Gordon,
`1984; Amanzio and Benedetti, 1999; Benedetti et al., 1999), which
`have subsequently been confirmed by brain imaging and in vivo
`receptor binding techniques (Petrovic et al., 2002; Zubieta et al.,
`2005).
`
`Materials and Methods
`Subjects. A total of 49 healthy volunteers participated in the study after
`they signed a written informed consent form in which the experimental
`procedure and the use of the different drugs were described in detail.
`Each subject underwent a complete clinical examination to rule out main
`diseases. To avoid high variability in hormonal responses, we adopted the
`following criteria. The subjects were tested with the State–Trait Anxiety
`Inventory (STAI) to rule out both trait and state anxiety. In the Italian
`population, the normal STAI-S range is 45.2 ⫾ 12.37 for adult women
`and 40.17 ⫾ 10.01 for men, whereas the STAI-T normal range is 46.1 ⫾
`11.53 for women and 39.53 ⫾ 9.25 for men (Spielberg et al., 1980; Nat-
`tero et al., 1989). In our routine clinical experience, these values corre-
`spond to a range of ⬃32–52 for the STAI-S and ⬃30 –55 for the STAI-T.
`Therefore, only subjects within these ranges were included in the study.
`In addition, we decided to include only those subjects within a prede-
`termined range of hormonal plasma concentrations. In the normal pop-
`ulation, the range of adrenocorticotropic hormone (ACTH) plasma con-
`centration in the morning is ⬃10 – 40 pg/ml, whereas the range of
`cortisol concentration is ⬃80 –300 ␮g/L (Liddle, 1974). To make the
`subjects as similar as possible, we included only those subjects who
`showed ACTH plasma concentration between 14 and 30 pg/ml and cor-
`tisol concentration between 90 and 130 ␮g/L.
`All of the experimental procedures were conducted in conformance
`with the policies and principles contained in the Declaration of Helsinki.
`The 49 subjects were randomly subdivided into four groups, whose char-
`acteristics are shown in Table 1.
`Pain induction. Pain was induced experimentally by means of the tour-
`niquet technique, according to the procedures described by Amanzio and
`Benedetti (1999). Briefly, the subject reclined on a bed, his or her non-
`dominant forearm was extended vertically, and venous blood was
`drained by means of an Esmarch bandage. A sphygmomanometer was
`placed around the upper arm and inflated to a pressure of 300 mmHg.
`The Esmarch bandage was maintained around the forearm, which was
`lowered on the subject’s side. After this, the subject started squeezing a
`hand spring exerciser 12 times while his or her arm rested on the bed.
`Each squeeze was timed to last 2 s, followed by a 2 srest. The force
`necessary to bring the handles together was 7.2 kg. This type of ischemic
`pain increases over time very quickly, and the pain becomes unbearable
`after ⬃14 min (Amanzio and Benedetti, 1999). The tourniquet test lasted
`10 min in all subjects, who had to rate their pain intensity every minute
`according to a numerical rating scale, ranging from 0 (no pain) to 10
`(unbearable pain).
`Drugs. Diazepam (Valium; Roche, Indianapolis, IN) was given intra-
`venously 30 min before the beginning of the tourniquet at a dose of 0.28
`mg/kg, with an infusion rate of 0.028 mg 䡠 kg ⫺1 䡠 min ⫺1 and a total
`
`infusion time of 10 min. Likewise, proglumide (Milid; Rottapharm, Mi-
`lan, Italy) was administered intravenously 30 min before the tourniquet
`at a dose of 1.5 mg/kg, with an infusion time of 0.15 mg 䡠 kg ⫺1 䡠 min ⫺1
`and a total infusion time of 10 min. As shown in Figure 1 A, proglumide
`is a glutamic acid-based CCK antagonist and is a nonselective antagonist
`that binds to both CCK-A and CCK-B receptors, or CCK-1 and CCK-2
`according to the new classification (Noble et al., 1999). Note that the
`binding affinity, expressed as the concentration required to inhibit by
`50% the specific binding of 125I-Bolton-Hunter CCK-8 (IC50), is similar
`for CCK-A and CCK-B receptors, although it is a little bit higher for
`CCK-A receptors (Benedetti, 1997). The time interval from the sphyg-
`momanometer cuff inflation (which lasted ⬃10 s) to the last squeeze was
`50 s, for a total of 1 min of inflation plus squeezing. Thus, the time
`interval from the end of drug administration to the last squeeze during
`the tourniquet was the same in all subjects (31 min). By considering that
`the tourniquet lasted 10 min in all subjects, the time interval from the end
`of drug administration to the end of the tourniquet was 41 min.
`Experimental design. The experiments were always performed at 9:00
`A.M. to avoid variability in the basal activity of the hypothalamic–pitu-
`itary–adrenal (HPA) axis, according to a randomized double-blind de-
`sign in which neither the subject nor the experimenter knew what drug
`was being administered. To do this, either the active drug or saline solu-
`tion was given. To avoid a large number of subjects, two or three addi-
`tional subjects per group received an infusion of saline in place of the
`active drug 30 min before the tourniquet. These subjects were not in-
`cluded in the study because they were used only to allow the double-blind
`design, as described previously by Benedetti et al. (2003).
`The complete experimental procedure is shown in Figure 1 B. Group 1
`(no-treatment or natural history group; n ⫽ 12) was tested twice (with an
`interval of 4 d) with the tourniquet without receiving any treatment.
`Likewise, group 2 (n ⫽ 13) was tested twice, with an intertest interval of
`4 d. In the first test, these subjects did not receive any treatment, whereas
`in the second test, they underwent a nocebo procedure. This consisted in
`the oral administration of an inert talc pill 5 min before the tourniquet,
`along with the verbal suggestions that it was a powerful vasoconstrictor
`further increasing the tourniquet-induced ischemia. The subjects were
`further told that, because of the quick vasoconstriction, this would in-
`duce a faster and larger increase of pain intensity, so that a quite strong
`hyperalgesic effect should be expected. To further strengthen the nocebo
`verbal suggestions, the subjects were told that they could give up at any
`time. Therefore, this experimental paradigm represents a situation in
`which a stressor is anticipated. Group 3 (n ⫽ 12) was tested twice (intert-
`est interval of 4 d), like group 2, but these subjects received a pretreat-
`ment with diazepam 30 min before the tourniquet. Similarly, group 4
`(n ⫽ 12) was tested with the tourniquet, like groups 2 and 3, but 30 min
`after a pretreatment with proglumide.
`ACTH and cortisol plasma concentration. Plasma concentrations of
`ACTH and cortisol were assessed before the tourniquet and at 5 and 10
`min during the tourniquet test (Fig. 1C), according to standard clinical
`practice and as described previously (Rainero et al., 2001; Benedetti et al.,
`2003). Briefly, blood samples were collected just before and at 5 and 10
`min of the tourniquet in sterile tubes and immediately centrifuged at 4°C,
`and the plasma was stored at ⫺80°C until assayed. Plasma ACTH and
`cortisol concentrations were measured using commercially available kit
`[ACTH Allegro (Nichols Institute, San Juan Capistrano, CA); CORT-
`CTK-125 (Sorin, Saluggia, Italy)]. The sensitivity of ACTH was 1 pg/ml,
`and the intraassay and interassay coefficient of variation were 3 and 7.8%.
`The sensitivity of cortisol was 5 ␮g/L, and the intraassay and interassay
`
`

`

`12016 • J. Neurosci., November 15, 2006 • 26(46):12014 –12022
`
`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`coefficients of variation were 3.8 and 5.7%. All of the samples from each
`subject were analyzed in the same assay.
`Statistical analysis. Because the experimental design involves both a
`between- and a within-subjects design, statistical analysis was performed
`by means of one-way ANOVA and ANOVA for repeated measures, fol-
`lowed by the post hoc Newman–Keuls test for multiple comparisons and
`Dunnett’s test for comparisons between a control group and different
`experimental groups. In addition, correlations were performed by using
`linear regression analysis. Data are presented as mean ⫾ SD, and the level
`of significance is p ⬍ 0.05.
`
`Results
`As shown in Table 1, no difference was present in the four differ-
`ent groups, for sex, age, weight, and basal plasma concentrations
`of ACTH and cortisol. In addition, STAI scores for both state and
`trait anxiety did not differ among the groups.
`The induction of ischemic pain in the no-treatment group
`(group 1) produced a type of pain that increased over time (Fig.
`2 A), along with increases in plasma concentrations of both
`ACTH and cortisol (Fig. 2 B, C, respectively). It is important to
`note that no difference was found between the first (black circles
`or squares) and second (white circles or squares) tests for all of
`these outcome measures (pain intensity, ACTH and cortisol).
`Therefore, our experimental conditions were stable in the two
`tests, because the repetition of the experimental pain and hor-
`monal assessment after 4 d did not produce different results.
`In group 2, the nocebo suggestions delivered on the second
`test (white circles or squares) induced both a significant increase
`of subjective pain rating (Fig. 3A) and hyperactivity of the HPA
`axis, as shown by the increased plasma concentrations of ACTH
`and cortisol (Fig. 3 B, C). Pain intensity increased in both the first
`⫽
`(black circles) and second (white circles) tests (F(19,228)
`133.855; p ⬍ 0.001), but the nocebo condition in the second test
`always induced higher pain scores compared with the first test, as
`assessed by the post hoc Newman–Keuls test for multiple compar-
`isons (Fig. 3A). In fact, at the end of the 10 min test, pain rating
`was 5.07 ⫾ 0.95 in the non-nocebo condition and 8.61 ⫾ 0.96 in
`⫽ 18.304; p ⬍ 0.001). Likewise,
`the nocebo condition (q(228)
`ACTH and cortisol increases occurred in both the first and sec-
`⫽
`⫽ 135.113, p ⬍ 0.001 for ACTH; F(5,60)
`ond tests (F(5,60)
`141.248, p ⬍ 0.001 for cortisol), but they were significantly larger
`in the nocebo condition (white squares) than in the first test
`(black squares), as assessed by means of the post hoc Newman–
`Keuls test (Fig. 3 B, C). In fact, at the end of the 10 min test, ACTH
`plasma concentration was 66.46 ⫾ 16.88 pg/ml in the non-
`nocebo condition and 86.69 ⫾ 8.45 pg/ml in the nocebo condi-
`⫽ 8.960; p ⬍ 0.01), and cortisol was 138.2 ⫾ 8.38 and
`tion (q(60)
`174.2 ⫾ 12.46 ␮g/L, respectively (q(60)
`⫽ 16.671; p ⬍ 0.001).
`The administration of diazepam blocked both the nocebo-
`induced hyperalgesia (Fig. 4 A) and the nocebo-induced hyper-
`activity of the HPA axis (Fig. 4 B, C). In fact, no difference was
`found between the first and the second tests, for both pain scores
`and hormone plasma concentrations. In contrast, proglumide
`only blocked the nocebo-induced hyperalgesia (Fig. 5A), but it
`was ineffective in reducing the nocebo-induced hyperactivity of
`the HPA axis (Fig. 5 B, C). The plasma concentrations of both
`⫽ 204.083,
`ACTH and cortisol increased in the two tests (F(5,55)
`p ⬍ 0.001 for ACTH; F(5,55)
`⫽ 147.952, p ⬍ 0.001 for cortisol),
`and the post hoc Newman–Keuls test showed a significantly
`higher hormone increase in the nocebo condition. In fact, at the
`end of the 10 min test, ACTH plasma concentration was 68.33 ⫾
`12.36 pg/ml in the non-nocebo condition and 87.5 ⫾ 6.26 pg/ml
`⫽ 10.302; p ⬍ 0.001), and cortisol
`in the nocebo condition (q(55)
`
`A, Proglumide is a nonselective CCK antagonist that binds to both CCK-A (or CCK-1)
`Figure 1.
`and CCK-B (or CCK-2) receptors. Note the similar binding affinity, expressed as the concentration
`required to inhibit by 50% the specific binding of 125I-Bolton-Hunter CCK-8 (IC50), for CCK-A and
`CCK-B receptors. B, Experimental design used in the present study. Each group was tested twice
`with tourniquet, and the interval between tests 1 and 2 was 4 d. C, Each tourniquet test lasted
`10 min, and the subjects had to rate their pain every minute while a blood sample was taken just
`before and at 5 and 10 min during the tourniquet.
`
`

`

`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`J. Neurosci., November 15, 2006 • 26(46):12014 –12022 • 12017
`
`The natural history of pain (A), ACTH (B), and cortisol (C) in test 1 (black symbols)
`Figure 2.
`is compared with the natural history of test 2 (white symbols). Note that no difference was
`present between tests 1 and 2, indicating stable conditions of our experimental setup.
`
`The natural history of pain (A), ACTH (B), and cortisol (C) in test 1 (black symbols) is
`Figure 3.
`compared with the nocebo condition of test 2 (white symbols), in which verbal suggestions of pain
`worsening were given. Note that, in the nocebo condition, there was a significant increase of
`both pain perception and ACTH and cortisol plasma concentrations. **p ⬍ 0.01; ***p ⬍ 0.001.
`
`

`

`12018 • J. Neurosci., November 15, 2006 • 26(46):12014 –12022
`
`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`⫽
`
`was 140.9 ⫾ 8.55 and 174.1 ⫾ 14.48 ␮g/L, respectively (q(55)
`15.262; p ⬍ 0.001).
`An intergroup one-way ANOVA showed that neither diaze-
`pam nor proglumide had an analgesic effect in non-nocebo con-
`ditions. In fact, there was no significant difference in pain time
`course between diazepam in the non-nocebo condition (Fig. 4 A,
`black circles) and the natural history of group 1 (Fig. 2 A). Like-
`wise, there was no difference in the time course of pain intensity
`between proglumide in the non-nocebo condition (Fig. 5A, black
`circles) and the natural history of group 1 (Fig. 2 A). In contrast,
`a significant decrease of pain occurred in the nocebo condition
`with both diazepam and proglumide compared with the drug-
`free nocebo condition of Figure 3A (white circles). In fact, at the
`end of the 10 min test, pain rating was 5.16 ⫾ 0.57 with diazepam
`and 5.58 ⫾ 0.9 with proglumide compared with 8.61 ⫾ 0.96 in
`⫽ 116.6, p ⬍ 0.001;
`the drug-free condition of group 2 (F(1,23)
`⫽ 66.07, p ⬍ 0.001, respectively). Therefore, whereas both
`F(1,23)
`diazepam and proglumide were ineffective as analgesics on basal
`pain, they proved to be effective in reducing the nocebo hyperal-
`gesic component. In other words, both drugs affected only the
`nocebo component of pain.
`The intergroup analysis of the hormonal responses produced
`similar results for diazepam but not for proglumide. In fact, there
`was no significant difference in ACTH– cortisol time course be-
`tween diazepam in the non-nocebo condition (Fig. 4 B, C, black
`squares) and the natural history of group 1 (Fig. 2 B, C). In con-
`trast, a significant decrease of both ACTH and cortisol plasma
`concentrations occurred in the nocebo condition with diazepam
`compared with the drug-free nocebo condition of Figure 3, B and
`C (white squares). In fact, at the end of the 10 min test, ACTH
`plasma concentration was 54.75 ⫾ 10.01 pg/ml with diazepam
`compared with 86.69 ⫾ 8.45 pg/ml in the drug-free condition of
`⫽ 74.74; p ⬍ 0.001), and cortisol concentration
`group 2 (F(1,23)
`was 142 ⫾ 8.31 ␮g/L compared with 174.2 ⫾ 12.46 ␮g/L in group
`⫽ 56.74; p ⬍ 0.001). In contrast to the hormonal effects
`2 (F(1,23)
`of diazepam, proglumide induced no effects on ACTH and cor-
`tisol, in neither the non-nocebo nor nocebo condition, as shown
`by no significant differences in the intergroup analysis between
`groups 1, 2, and 4.
`Finally, a correlation analysis between pain ratings and hor-
`mone plasma concentrations did not show any significant effect
`in the non-nocebo and the nocebo conditions.
`
`Discussion
`Most of placebo research in recent times has focused on placebo
`effects in pain and Parkinson’s disease. In the first case, there are
`now several converging lines of evidence indicating that placebo-
`induced expectations of analgesia activate the endogenous opioid
`systems in some circumstances (Benedetti et al., 2005; Colloca
`and Benedetti, 2005; Hoffman et al., 2005; Zubieta et al., 2005). In
`the second case, dopamine release in the striatum seems to play
`an important role (de la Fuente-Fernandez et al., 2001), and the
`placebo response in Parkinson patients is associated with neuro-
`nal changes in the subthalamic nucleus (Benedetti et al., 2004).
`In contrast, the study of the nocebo effect has not been care-
`fully investigated, although some attempts to analyze its under-
`lying neurobiological mechanisms have been performed
`(Benedetti et al., 1997, 2003; Johansen et al., 2003). For example,
`we tried to assess the role of CCK in nocebo hyperalgesia in
`postoperative patients by using a neuropharmacological ap-
`proach with the CCK antagonist proglumide (Benedetti et al.,
`1997). Although in our previous study we showed a blockade of
`nocebo hyperalgesia by proglumide, the clinical experimental
`
`The effect of diazepam on basal pain increase (A), basal ACTH increase (B), and
`Figure 4.
`basal cortisol increase (C) in test 1 (black symbols) is compared with diazepam in the nocebo
`condition of test 2 (white symbols). Note that diazepam suppressed both nocebo hyperalgesia
`and nocebo increase of ACTH and cortisol plasma concentrations. In fact, no difference was
`present between tests 1 and 2.
`
`

`

`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`J. Neurosci., November 15, 2006 • 26(46):12014 –12022 • 12019
`
`setting presented many ethical limitations. Therefore, in the
`present study, we addressed several unanswered questions of the
`study by Benedetti et al. (1997) by performing a careful analysis of
`the effects of proglumide on both pain and HPA axis in different
`nocebo and non-nocebo conditions and, moreover, by compar-
`ing the effects of proglumide with those of diazepam, a widely
`known anti-anxiety drug.
`First of all, some methodological considerations and some
`limitations of the present study are worthy of discussion. We
`studied ACTH and cortisol because several studies have shown
`that the plasma concentrations of these hormones are sensitive to
`a number of stressors (Dickerson and Kemeny, 2004), including
`experimental ischemic arm pain (Gullner et al., 1982; Johansen et
`al., 2003). In particular, a short-latency response of cortisol has
`been shown in ischemic pain (Johansen et al., 2003), thus making
`our 10-min-long experimental pain a good model for HPA anal-
`ysis. We used a neuropharmacological approach in humans with
`a nonspecific CCK antagonist for at least two reasons. First, the
`neuropharmacological approach to placebo analgesia with the
`opioid antagonist naloxone has been crucial to the understanding
`of the neurobiology of the placebo analgesic effect (Levine et al.,
`1978; Grevert et al., 1983; Levine and Gordon, 1984; Benedetti,
`1996; Amanzio and Benedetti, 1999; Benedetti et al., 1999). In
`fact, these early pharmacological studies have been confirmed
`recently by several brain imaging studies, which basically show a
`similarity between narcotics and placebos in the activation of
`different brain regions (Petrovic et al., 2002) and the in vivo acti-
`vation of the ␮-opioid receptors following a placebo procedure
`(Zubieta et al., 2005). Therefore, the neuropharmacological
`study of placebo/nocebo phenomena with agonist and antagonist
`drugs in humans appears to be a reliable experimental approach
`that gives important information. The second reason why we
`decided to use the pharmacological approach with proglumide is
`that we were not so much interested in the involvement of spe-
`cific CCK receptors but rather in studying CCK from a general
`point of view. In this sense, proglumide has proven to be useful in
`investigating placebo analgesia (Benedetti et al., 1995; Benedetti,
`1996; Colloca and Benedetti, 2005).
`Although proglumide is a weak CCK antagonist, its anti-CCK
`action in the brain has been demonstrated. There is behavioral
`and electrophysiological evidence that CCK is blocked by proglu-
`mide in the brain (Chiodo and Bunney, 1983; Suberg et al., 1985;
`Watkins et al., 1985a,b). The results obtained in humans on opi-
`oid potentiation by proglumide (Price et al., 1985; Lavigne et al.,
`1989; Benedetti et al., 1995; Benedetti, 1996) are in keeping with
`the potentiation of morphine analgesia by the CCK-A antagonist
`devazepide in the rat (Dourish et al., 1988) and with the results
`obtained in animal
`studies using CCK-B antagonists
`(Wiesenfeld-Hallin et al., 1990; Maldonado et al., 1993; Noble et
`al., 1993; Valverde et al., 1994; Xu et al., 1994; Andre et al., 2005).
`Proglumide has also been reported to block the anxiogenic effects
`of the tetrapeptide CCK-4 and caerulein, a CCK-8 agonist, indi-
`cating an anti-CCK action in the CNS at the level of affective
`mechanisms (Harro et al., 1990; Harro and Vasar, 1991; Van
`Megen et al., 1994).
`By taking into consideration the limitations discussed above,
`the present study suggests that the nocebo hyperalgesic counter-
`part of the placebo/nocebo phenomenon is mediated by CCK. In
`particular, it suggests that the CCK antagonist proglumide does
`not act on the nocebo-induced anxiety but rather on anxiety-
`induced hyperalgesia. In fact, whereas diazepam reduced both
`HPA activation and pain perception, proglumide affected pain
`but not the HPA axis. The most plausible explanation of our
`
`The effect of proglumide on basal pain increase (A), basal ACTH increase (B), and
`Figure 5.
`basal cortisol increase (C) in test 1 (black symbols) is compared with proglumide in the nocebo
`condition of test 2 (white symbols). Note that, whereas proglumide suppressed nocebo hyper-
`algesia, it was ineffective in suppressing the nocebo increase of ACTH and cortisol plasma con-
`centrations. **p ⬍ 0.01; ***p ⬍ 0.001.
`
`

`

`12020 • J. Neurosci., November 15, 2006 • 26(46):12014 –12022
`
`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`Figure 6. Model to explain the findings of the present study. Nocebo-induced anxiety affects
`both the HPA axis and pain mechanisms. The link between anxiety and pain is represented by
`CCK, which has a facilitating effect on pain. Benzodiazepines, like diazepam, can block anxiety,
`thus preventing both HPA hyperactivity and hyperalgesia. CCK antagonists, such as proglumide,
`only block the CCKergic anxiety–pain link. Therefore, CCK antagonists do not inhibit pain per se
`but rather the anxiety–pain link.
`
`findings is shown in Figure 6. The reduction of both pain and
`HPA hyperactivity by diazepam can be explained by its anxiolytic
`effect, thus affecting nocebo-induced anxiety. It should be noted,
`however, that nonspecific effects of diazepam, e.g., on arousal,
`cannot be ruled out completely. In this regard, it will be interest-
`ing to use measures of anxiety in future studies. In contrast, pro-
`glumide was likely to affect only the CCK-mediated link between
`anxiety and pain. Although we did not test naloxone in the
`present study, the involvement of endogenous opioids in the
`blockade of nocebo hyperalgesia seems to be unlikely, as shown
`by the ineffectiveness of naloxone in our previous study
`(Benedetti et al., 1997). This CCK link between anxiety and pain
`is in agreement with studies in rodents in which more selective
`CCK antagonists were used. For example, a recent study showed
`that CI-988, a specific CCK-B receptor antagonist, blocked
`anxiety-induced hyperalgesia, which indicates a biochemical link
`between anxiety and pain that is mediated by CCK-B receptors
`(Andre et al., 2005).
`In recent years, there has been accumulating evidence that
`CCK acts as a neuromodulator of different functions, such as
`pain and anxiety, although the exact mechanisms are still unclear.
`CCK is found in the brain as an octapeptide (CCK-8) and has
`
`Figure 7.
`The structural formula of the anxiolytic drug diazepam, one of the most used
`benzodiazepines, is similar to L-365,260, a benzodiazepine-based CCK antagonist, thus sug-
`gesting similar mechanisms of action of benzodiazepines and CCK antagonists. Note the higher
`binding affinity of L-365,260 for CCK-B receptors compared with CCK-A receptors, expressed as
`the concentration required to inhibit by 50% the specific binding of 125I-Bolton-Hunter CCK-8
`(IC50 ).
`
`

`

`Benedetti et al. • The Hyperalgesic Nocebo Effect
`
`J. Neurosci., November 15, 2006 • 26(46):12014 –12022 • 12021
`
`many functions, ranging from pain modulation to anxiety
`(Vanderhaegen et al., 1975; Beinfeld, 1983; Baber et al., 1989;
`Crawley and Corwin, 1994; Hebb et al., 2005). The distribution of
`CCK in the brain matches that of the opioid peptides at both the
`spinal and supraspinal level (Stengaard-Pedersen and Larsson,
`1981; Gall et al., 1987; Gibbins et al., 1987), suggesting a close
`interaction between the two neuropeptides.
`The involvement of CCK in both pain modulation and anxi-
`ety is particularly relevant to the present study. Interestingly, it is
`worth noting that some CCK-B receptor antagonists, such as
`[3S(⫺)[N⬘-2,3-dehydro-1-methyl-2-oxo5-phenyl-
`L-365,260
`1 H-1,4-benzodiazepin-3-yl]-1 H-indole-2-carboxamide], have a
`benzodiazepine-based chemical structure that is similar to the
`anxiolytic drug diazepam (Fig. 7), thus suggesting a similarity of
`action of CCK antagonists and benzodiazepines. In this regard,
`the recent work by Andre et al. (2005) in the rat shows that
`CCK-B receptor blockade antagonizes anxiety-induced hyperal-
`gesia. This suggests a CCKergic link between anxiety and hyper-
`algesia, whereby anxiety-activated CCK has a facilitating action
`on pain. Our present study confirms these effects in humans and
`suggests that our nocebo procedure induced anticipatory anxiety
`about the impending pain.
`It should also be stressed that the anti-CCK action of proglu-
`mide did not show any real analgesic effect, because the basal pain
`increase of the natural history group was unaffected. In other
`words, proglumide was effective only on the nocebo component
`of pain, that is, only on the anxiety-induced hyperalgesia. There-
`fore, CCK antagonists appear to be useful not so much as anal-
`gesics but rather as drugs suppressing the hyperalgesia induc