Placebo and Opioid Analgesia-- Imaging a Shared Neuronal Network

See allHide authors and affiliations

Science  01 Mar 2002:
Vol. 295, Issue 5560, pp. 1737-1740
DOI: 10.1126/science.1067176


It has been suggested that placebo analgesia involves both higher order cognitive networks and endogenous opioid systems. The rostral anterior cingulate cortex (rACC) and the brainstem are implicated in opioid analgesia, suggesting a similar role for these structures in placebo analgesia. Using positron emission tomography, we confirmed that both opioid and placebo analgesia are associated with increased activity in the rACC. We also observed a covariation between the activity in the rACC and the brainstem during both opioid and placebo analgesia, but not during the pain-only condition. These findings indicate a related neural mechanism in placebo and opioid analgesia.

Placebo analgesia is an important component in pain management (1), although the basic mechanisms are still poorly understood. At least some aspects of placebo analgesia are dependent upon endogenous opioid systems (1–3) because the effect may be partly abolished by the opioid antagonist naloxone (2). Therefore, the underlying neurophysiology of opioid-dependent placebo analgesia can be elucidated by studying similarities and differences in the function of the opioid and placebo systems in the brain. The opioid system consists of a well-studied subsystem in the brainstem and a less well elaborated cortical opioid-dependent network (4, 5). This system appears to be a likely candidate for the mediation of opioid-dependent placebo analgesia. The importance of the ACC in opioid effects has been suggested in several receptor-imaging studies of the brain (6–10), activation studies of opioid compounds (11–14), and theoretical frameworks of opioid analgesia (15). The rostral ACC (rACC)/ventromedial prefrontal cortex has been suggested as an important region in opioid analgesia and in other forms of pain modulation (16–26), which may suggest a similar involvement of higher order control of opioid-dependent placebo analgesia.

We compared the analgesic effects of a placebo treatment and a rapidly acting opioid (remifentanil) [supplement A (27)] in a standard pain-stimulus paradigm (28). We used six different conditions in the study: heat pain and opioid treatment (POP), nonpainful warm stimulation and opioid treatment (WOP), heat pain and placebo treatment (PPL), nonpainful warm stimulation and placebo treatment (WPL), heat pain only (P), and nonpainful warm stimulation only (W). We studied concomitant behavioral responses and regional cerebral blood flow (rCBF) using positron emission tomography (PET) (29, 30) and compared the functional anatomy of the placebo analgesic response with that of the opioid response. We were especially interested in whether placebo analgesia and opioid effects induce a similar rCBF response in the rACC and the brainstem.

Comparison of scans in the pain conditions and in the warm conditions showed increased activity in the contralateral thalamus, in the insula bilaterally, and in the caudal ACC [Web table 1 (27) and Fig. 1A], all regions that have shown increased activity in previous imaging studies of pain (31). The opioid agonist remifentanil activated the cerebral network [Web table 2 (27) and Figs. 1B and 2A], which has been described previously in opioid receptor binding (6–9) and in rCBF studies (11–14). One of the major increases in rCBF was observed in the ACC and especially in the rACC. We also observed an increased activity in the lower brainstem. The subjects rated the pain intensity lower during POP compared with P in every experimental block [Web fig. 1 (27)]. The rCBF analysis showed that the insula, one of the major regions involved in pain processing, had an attenuated rCBF response bilaterally during POP-WOP as compared with P-W [Web table 2 (27)].

Figure 1

(A) Increased activity was observed in the right (cross) and left insula (left panel, horizontal section), in the thalamus (left panel), and in the caudal ACC (right panel, sagittal section) during the main effect of pain [(POP + PPL + P)-(WOP + WPL + W)]. (B) The activation was most pronounced in the rACC during the main effect of opioids [(POP + WOP)-(P + W)]. Increased activity is apparent in the lower pons. (C) Increased activity in the same area of the rACC was also seen in the placebo effect during pain (PPL-P). The activations are presented on an SPM99-template. The activation threshold is at P= 0.005.

Figure 2

(A) The main effect of remifentanil [(POP + WOP)-(P + W)] showed increased activity bilaterally in the rostral and caudal ACC (extending into the ventromedial prefrontal cortex), insula, orbitofrontal cortex (extending into the temporopolar areas), and lower pons. The effect was widespread in the rACC and bilaterally in the anterior insula. (B) The placebo effect during pain (PPL-P) showed increased activity in the orbitofrontal regions bilaterally (most extensively in the right hemisphere) and in the contralateral rACC. (C) To observe the overlapping activation in the two different conditions, we used the placebo analgesia effect (activation threshold at P = 0.001) and masked the main effect of remifentanil (same activation threshold). Several of the orbitofron- tal regions in the right hemisphere and in the rACC remained after the high-threshold masking, indicating that these regions were activated both during opioid stimulation and during the pain and placebo conditions. Thus, these regions were activated both by opioids in general and by placebo during pain. The activations are presented on an SPM99-template. The activation threshold is at P = 0.005.

Although there was high interindividual variability in placebo ratings, most subjects decreased their pain intensity rating during PPL as compared with the P condition [Web fig. 1 (27)]. Recent experiments have revealed different types of placebo analgesia and indicate that some are dependent upon opioid systems (32, 33). Placebo responses were induced in subjects as a result of suggestions that each of the drugs used in the experiment was a potent analgesic (28) (i.e., expectation of pain relief) and by preceding the placebo treatment by active opioid during noxious stimulation in the first (five subjects) or second experimental block (four subjects) (i.e., opioid conditioning). Both of these placebo mechanisms can be abolished by the opioid antagonist naloxone and thus appear to be opioid dependent (32). Therefore, we expected similarities between activity observed in the opioid network and in the placebo analgesia network. The placebo analgesia was accompanied by increased activity in the orbitofrontal and ACC areas during PPL when compared with P [Web table 3 (27); Figs. 1C and 2B]. When we controlled for unspecific placebo effects (WPL-W), we observed an activation in the ACC [Web table 3 (27)], somewhat caudal to the rACC effect in PPL-P but rostral to the ACC activation during pain.

Previous imaging studies have shown that the rACC is more reliably activated by opioids, whereas the caudal ACC is more reliably activated by pain (11, 12). This distinction was also observed here, pointing to the importance of the rACC in opioid analgesia. This area of the human ACC contains a high concentration of opioid receptors (9). Moreover, studies examining stimulus-induced analgesia (16, 18,19, 21, 23), nitrous oxide–induced analgesia (22), and hypnosis-induced change in pain perception (24, 25) have shown an increased activation in the rACC associated with the modulation mechanism. Rainville et al. (25) noted a similar functional division of the ACC: Pain (and unpleasantness) activated a more caudal region in the ACC, whereas the conditions involving suggestion, resulting in modulation of the pain experience, activated a more rostral part in the ACC. Hence, the increased activity in the rACC during PPL-P may support its involvement in the analgesic response mechanism during placebo. In addition, a post hoc analysis indicated that during opioid analgesia, the high placebo responders activated this area, whereas the low responders did not (Fig. 3). This suggests a relation between how effectively opioids may activate the rACC and adjacent areas and how well subjects respond to placebo during pain. Earlier studies have shown a behavioral correlation between opioid analgesia and placebo analgesia (32). The suggestion that the opioid system may vary among subjects is supported by the finding that the opioid receptor binding potential during rest and pain is highly specific to an individual (10), leading to the hypothesis that high placebo responders have a more efficient opioid system.

Figure 3

Post hoc analysis comparing the activation of the rACC in high placebo responders with that in low responders revealed no significant differences in PPL-P between groups. However, activation of the rACC and adjacent areas by the high placebo responders was significant during POP-P [(x, y,z) = (−2, 46, 22); Z = 4.77] (A), whereas activation by the low responders was not significant (B). The difference between groups was significant in the rACC/ventromedial prefrontal cortex [(x,y, z) = (−2, 46, 26); Z = 3.24]. The activation threshold is at P = 0.005.

The placebo analgesic effect is dependent on complex cognitive information processing, including analysis of threat in a given context, expectations of treatment outcome, and desire for relief (1, 3, 4). The brainstem opioid system may thus be under cognitive control from higher order cortical regions. The ACC might play a key role in the cortical control of the brainstem during opioid analgesia (15,34) by way of fiber tracts projecting directly to the periaqueductal gray (PAG) (35) or by way of the medial thalamic nucleus (36). A similar mechanism may be necessary in placebo analgesia, which implies that a functional connection should exist between these regions, both in opioid and placebo analgesia. Regression analysis supported this hypothesis (30) [Web table 4 (27) and Fig. 4]. The activity in the rACC covaried with activity in areas close to the PAG and the pons in the POP condition. We also observed a significant covariation in activity between the rACC and the pons, and a subsignificant covariation in activity between the rACC and the PAG, during PPL. No effect was observed in the pain-only condition (P), and the differences between these regressions (POP versus P and PPL versus P) were significant. The area in the pons is in the same region as the area activated in the main effect of opioids (Fig. 1). The brainstem opioid system consists of the PAG, which alters the neuronal activity in the rostral ventromedial medulla (4, 5). Additional nuclei in the pons, such as the parabrachial nuclei, also contain opioid-dependent neurons (4, 5). The positive covariation between rACC and these regions during POP and PPL, but not during P, may thus indicate that the higher cortical systems may, in specific circumstances, exert direct control over the analgesic systems of the brainstem not only during opioid analgesia but also during placebo analgesia.

Figure 4

(A to C) Covariation of activity in brainstem regions with activity in the rACC (denoted by the blue sphere) in different pain conditions. (A) Activity in the rACC covaried with activity in the PAG and in the lower pons/medulla during the POP condition. These covariations were significantly greater during POP as compared with P [Web table 4 (27)]. (B) A similar covariation between the rACC and the lower pons/medulla was observed during the PPL condition. This covariation was significantly greater during PPL as compared with P [Web table 4 (27)]. (C) No such regressions were observed during the P condition. The activations are presented on an SPM99-template and a more detailed image of the brainstem indicating the approximate position of the PAG and the pons. The threshold of activation is at P = 0.005.

The increased activity in the lateral orbitofrontal cortex during placebo analgesia is of interest because previous PET studies have implicated this region in cognitively driven pain modulation (25, 37). Stimulation of this region in rats (38) and primates (39) also induces analgesia. A right predominance of the orbitofrontal activation was observed during placebo analgesia, but interpretation of this finding is uncertain because it may reflect a threshold effect. Placebo analgesia seems to activate a more rostral part of the orbitofrontal cortex as compared with the general opioid effect. Because the orbitofrontal cortex has dense connections with both the ACC and the brainstem (40), which have also been implicated in placebo analgesia, we suggest that these regions belong to a network that uses cognitive cues to activate the endogenous opioid system.

  • * To whom correspondence should be addressed. E-mail: martin{at}


View Abstract

Navigate This Article