Placebo and Nocebo Effects in Clinical Trials and Clinical Practice

The Magnitude of Placebo Analgesia

Pain is by far the most studied placebo condition, and a large number of studies on the placebo effect have been performed in experimentally induced pain in healthy subjects or patients experiencing a painful condition. There are at least two explanations for this finding. First, pain is a subjective experience that undergoes psychological and social modulation more than any other condition. The fine tuning of the global pain experience by many psychosocial factors makes pain an excellent model for identifying and understanding the placebo effect. The second explanation stems from the influential work of Beecher in the 1950s. In fact, in 1955, Beecher reviewed 15 controlled trials involving 1802 patients. Defining positive outcomes as the “percent satisfactorily relieved by placebo,” Beecher reported effect sizes ranging from 26% to 58%, with an average of 35%. The notion that about one-third of patients respond to a placebo has since permeated medical texts and teachings, even though this work has been criticized on methodological grounds. ^, In fact, a real figure of Beecher’s study was that the magnitude of placebo analgesia might range from no responses to large responses. For example, Levine et al. found that 39% of patients had an analgesic response to placebo treatment, and in a study of ischemic arm pain in normal volunteers, Benedetti found that 26.9% of them responded to a placebo analgesic compared to a no treatment control group. Another study involving cutaneous heating of the left hand found that 56% of subjects responded to placebo treatment compared to the no treatment controls.

Assessing the magnitude of the placebo analgesic effect is not an easy task, as the experimental conditions and the psychological state of the subjects change across different studies. Several studies measured the average change in pain experienced by those receiving placebo and compared this figure to the average change in the no treatment group, and found that the magnitude of the placebo analgesic effect is about two out of ten units on a visual analog (VAS) or numeric rating scale (NRS). In studies where the known placebo responders in a group are separated for analysis, the average magnitude of analgesia has been found, not surprisingly, to be significantly greater. For example, when Benedetti (1996) looked only at responders, he found an average placebo analgesia magnitude of five units on the ten unit NRS. This is similar to a postoperative dental study that found a 3.3 cm (out of ten) lower mean post-treatment VAS score for placebo responders compared to nonresponders.

The experimental manipulation used to induce placebo analgesia played a fundamental role in the magnitude of the response. Verbal suggestions that induce certain expectations of analgesia induce larger placebo analgesic responses than those that induce uncertain expectations. This is illustrated by a study carried out in a clinical setting that investigated the differences between the double-blind and deceptive paradigms. Postoperative patients were treated with buprenorphine, on request, for three consecutive days consecutively and with a basal infusion of saline solution. However, the symbolic meaning of this saline basal infusion was varied in three different groups of patients: the first group (natural history or no-treatment group) was told nothing; the second group was told that the infusion was either a potent analgesic or a placebo (classic double-blind administration); and the third was told that the infusion was a potent painkiller (deceptive administration). The placebo effect of the infusion was measured by recording the doses of buprenorphine requested over the three day treatment. It is important to stress once again that the double-blind group received uncertain verbal instructions (“It can be either a placebo or a painkiller. Thus, we are not certain that the pain will subside”), whereas the deceptive administration group received certain instructions (‘It is a painkiller. Thus the pain will subside soon. Compared to the natural history group, a 20.8% decrease in buprenorphine intake was observed with double-blind administration. An even greater decrease (33.8%) was observed in the deceptive administration group. It is important to point out that the time course of pain was the same in all three groups over the three day treatment period. The same analgesic effect was observed with different doses of buprenorphine. Thus subtle differences in the verbal context of the patient may have a significant impact on the magnitude of the response. Verne et al. and Vase et al. conducted two similar studies in which patients with irritable bowel syndrome (IBS) were exposed to rectal distension using a balloon barostat, a type of visceral stimulation that simulates their clinical pain. They tested patients with untreated natural history (baseline), rectal placebo, and rectal lidocaine. Pain was rated immediately after each stimulus in each condition. The first study was conducted as a double-blind crossover clinical trial, so patients were given an informed consent form stating that they “may receive an active pain reducing medication or an inert placebo agent.” In this study, there was a significant pain-relieving effect of rectal lidocaine compared to rectal placebo, and a significant pain-relieving effect of rectal placebo compared to the no-treatment condition. In the second study, at the onset of each treatment condition (rectal placebo, rectal lidocaine), patients were told, “The agent you have just been given is known to significantly reduce pain in some patients.” A much larger placebo analgesic effect was found, which did not significantly differ from that of rectal lidocaine. These two studies show that adding an overt suggestion for pain relief can increase placebo analgesia to a magnitude that matches that of an active agent.

Experience can also influence the magnitude of placebo analgesia. In one study, the intensity of painful stimulation was reduced surreptitiously after placebo administration, leading subjects to believe that an analgesic treatment was effective. This procedure induced strong placebo responses after minutes, and these responses, albeit reduced, lasted up to four to seven days. This procedure was repeated four–seven days after totally ineffective analgesic treatment in the second group of participants. The placebo responses were markedly reduced compared to those in the first group. These results emphasize that the placebo effect may represent a learning phenomenon involving several factors and may explain the large variability in the magnitude of placebo responses among studies. The number of learning trials carried out in a conditioning procedure is also related to the subsequent magnitude of the placebo effects. Learning processes are not only important for psychophysical measurements such as pain rating, but also affect neurophysiologic parameters, such as laser-evoked potentials. Furthermore, and Colloca and Benedetti showed that social observational learning, whereby placebo responses can be learned by merely observing others, plays an important role.

As the magnitude of placebo analgesia depends on many factors, and learning is crucial in several circumstances, the clinical trial setting does not seem to be the best model for studying the placebo analgesic response. In the study by Hrobjartsson and Goetzsche, a meta-analysis was performed of 130 trials in which placebo and no treatment groups were identified in different pathologic conditions. No difference was found between the two groups for many conditions, but a significant placebo effect was found for pain in 29 clinical trials, indicating that pain is one of the best conditions where the placebo effect is present and can be studied. To further investigate the placebo effect in analgesic studies only, Vase et al. conducted two meta-analyses: one covered 23 of the 29 clinical trials included in Hrobjartsson and Goetzsche’s meta-analysis, and the other covered only 14 studies investigating the mechanisms of placebo analgesia. They found that the magnitude of placebo analgesic effects was higher in studies that investigated placebo analgesic mechanisms than in clinical trials in which the placebo was used only as a control condition.

Modulation of pain perception by placebos critically depends on expectation, as shown in many studies ; for example, Price et al. applied placebo creams and graded levels of heat stimulation on three adjacent cutaneous regions of the forearm, suggesting that cream A was a strong analgesic, cream B was a weak analgesic, and cream C was a control agent. Immediately after these conditioning trials, participants rated their expected pain levels for the placebo test trials, wherein the stimulus intensity was the same for all three regions. The conditioning trials led to graded levels of expected pain (C ≥ B ≥ A) for the three creams and graded magnitudes of actual pain (C ≥ B ≥ A) when tested during placebo test trials. Thus the magnitude of placebo analgesia could be graded across three adjacent skin areas, demonstrating a high degree of somatotopic specificity for placebo analgesia. Expected pain levels accounted for 25% to 36% of the variance in post-manipulation pain ratings. In another study, one group of subjects was pharmacologically pre-conditioned with ketorolac (a non-opioid analgesic) for two days in a row. The ketorolac was then replaced with a placebo on the third day, along with verbal suggestions for analgesia. This procedure induces a strong placebo analgesic response. To determine whether this response was because of pharmacologic pre-conditioning, the second group of subjects was also pre-conditioned with ketorolac, but the placebo was administered on the third day along with verbal suggestions that the drug was a hyperalgesic agent. These verbal instructions were sufficient not only to completely block the placebo analgesia but also to produce hyperalgesia. This study clearly shows that placebo analgesia depends on the expectation of a decrease in pain, even when analgesic preconditioning is performed.

The expectation of analgesia can also make a big difference in clinical trials, so the patient’s expectations must be considered whenever a clinical trial is conducted. In one clinical trial, real acupuncture was compared with sham acupuncture. Patients were asked which group they believed they belonged to (either placebo or real treatment). Patients who believed they belonged to the real treatment group experienced larger clinical improvement than those who believed they belonged to the placebo group. while in another clinical trial, patients with higher expectations about acupuncture experienced greater clinical benefits than those with lower expectations, regardless of their allocation to real or sham groups. It did not matter whether the patients actually received the real or the sham procedure—what mattered was whether they believed in acupuncture and expected a benefit from it.

Some of the best evidence underscoring the crucial role of positive and negative expectations in the outcome of analgesic treatments is the decreased effectiveness of hidden treatments. This involves giving an analgesic covertly (unexpectedly), so the patient is unaware a drug is being injected; the outcome following the hidden (unexpected) administration is compared with that following an open (expected) administration. ^, Careful analysis of the differences between open (expected) and hidden (unexpected) injections in postoperative settings was performed for five widely used painkillers (morphine, buprenorphine, tramadol, ketorolac, metamizole). ^,

That the hidden administration of a pharmacologic agent is less effective than an open one may suggest a different pharmacodynamic action of the drug in the absence of expectations. Although we do not know whether expectations and therapeutic rituals can modify a receptor to change the drug-receptor binding properties, to the best of our knowledge, this mechanism seems unlikely. The global effect of a drug derives from its specific pharmacodynamic action plus the psychological (placebo) effect coming from the very act of its administration. A recent study suggested that these two components operate independently from each other.

Expectations may also be associated with other factors, such as the desire for relief and reduction of anxiety. In other words, placebo phenomena occur within the context of emotional regulation, and symptoms should be influenced by desire, expectation, and intensity of emotional feeling. Desire and expectation interact and underlie common human emotions like sadness, anxiety, and relief, thus in the context of analgesic studies, it is quite plausible that patients and study participants have some degree of desire to avoid, terminate, or reduce evoked or ongoing pain.

Placebo-Induced Activation of Endogenous Painkillers

( Fig. 17.1 )

Placebo administration has been found to activate endogenous mechanisms of analgesia, and the placebo effect represents one of the most interesting models for understanding when and how these mechanisms are activated. The organization of the endogenous antinociceptive systems is complex. Thanks to the pioneering work on producing analgesia in rats by stimulation of the periaqueductal grey and the discovery of stereospecific binding sites for opioids, endogenous enkephalins in the central nervous system, we know that a complex endogenous network exists, which includes both opioid and non-opioid systems. ^,

Most of our knowledge about placebo analgesia is based on the opioid mechanisms. Opioid receptors can be found throughout the brain, brainstem, and spinal cord. These receptors may exert analgesic effects through different mechanisms, such as modulation at the spinal level and/or control of cortical and brainstem regions. The modulation of the spinal cord is one of the best described. ^, The opioid system in the brainstem consists of different regions, such as the periaqueductal gray, parabrachial nuclei, and the rostral ventromedial medulla. ^, Although opioid receptors are less characterized in the cortex, autoradiographic studies indicate high concentrations of opioid receptors in the cingulate cortex and prefrontal cortex, and one of the highest levels of opioid receptor binding has been found in the anterior cingulate cortex. Studies performed using positron emission tomography, and the radioactive opioid 11C-diprenorphine confirms previous animal and human autoradiographic findings. ^, In addition, opioid receptor agonists, such as remifentanil and fentanyl, have been shown to act on several regions that are known to be involved in pain processing and contain high concentrations of opioid receptors. ^,

One of the main problems of this endogenous opioid network is understanding when and how it is involved in analgesic effects. The first study of the biologic mechanisms of placebo analgesia used the opioid antagonist naloxone to block opioid receptors. It was performed in a clinical setting in patients who had undergone extraction of their third molar tooth. The investigators found disruption of placebo analgesia after naloxone administration, which indicates the involvement of endogenous opioid systems in the placebo analgesic effect. Since the publication of the study by Levine et al. attempts have been made to verify and reproduce the findings. For example, by using the experimental pain model of arm ischemia (the tourniquet technique), Grevert et al. found a partial reversal of placebo analgesia by naloxone, thus confirming previous findings. Likewise, Levine and Gordon adopted an elegant experimental design that showed naloxone-reversed placebo analgesia in a clinical situation, namely postoperative pain after extraction of the third molar. However, in 1983, Gracely et al. demonstrated that naloxone might have hyperalgesic effects on postoperative pain, thus shedding some doubt on the opioid hypothesis of placebo analgesia.

From 1995 to 1999, a long series of experiments with rigorous experimental designs were performed by Benedetti et al. Many unanswered questions have been clarified during these years, and the role of endogenous opioids in placebo analgesia has been explained. The experimental ischemic arm pain model was used to show that naloxone does not affect this type of pain. Therefore any effect following naloxone administration could be attributed to the blockade of placebo induced opioid activation. At the same time, these investigators tested the effects of a cholecystokinin (CCK)-antagonist, proglumide, on placebo analgesia. Based on the anti-opioid action of CCK, it was hypothesized that blockade of CCK receptors would enhance the opioids released by the placebo. Indeed, it was found that proglumide potentiated the placebo analgesia and presented a novel and indirect way to test the opioid hypothesis. ^, More recently, the activation of the CCK type-2 receptors was performed using the agonist pentagastrin, which disrupted placebo responses completely. Therefore the activation of the CCK type-2 receptors has the same effect as the μ-opioid receptor antagonist naloxone, suggesting that the balance between cholecystokinergic and opioidergic systems is crucial for placebo responsiveness in pain. Therefore one of the most interesting models involves two opposing neurotransmitter systems, opioids and CCK.

Specific placebo analgesic responses can be obtained in different parts of the body. ^, It was found that these responses are naloxone reversible. If four noxious stimuli are applied to the hands and feet, and a placebo cream is applied to one hand only, then pain is reduced only on the hand on which the placebo cream had been applied. This highly specific effect is blocked by naloxone, suggesting that the placebo-activated endogenous opioid systems have a precise and somatotopic organization.

Petrovic et al. found that some brain regions in the cerebral cortex and brainstem are affected by both a placebo and the rapidly acting opioid agonist remifentanil, indicating a related mechanism in placebo induced and opioid-induced analgesia. Administration of a placebo induced activation of the rostral anterior cingulate cortex and orbitofrontal cortex. Moreover, there was significant co-variation in activity between the rostral anterior cingulate cortex and the lower pons and medulla, and sub-significant co-variation in activity between the rostral anterior cingulate cortex and the periaqueductal gray. This suggested that the pain modulating circuit of the descending rostral anterior cingulate-periaqueductal gray-rostral ventromedial medulla is involved in placebo analgesia, as previously hypothesized by Fields and Price. The first direct evidence of opioid-mediated placebo analgesia was published in 2005. In vivo receptor binding techniques using the radiotracer carfentanil, a µ-opioid agonist, were used to show that a placebo procedure activates µ-opioid neurotransmission in the dorsolateral prefrontal cortex, the anterior cingulate cortex, the insula, and the nucleus accumbens. A more detailed account of µ-opioid neurotransmission after placebo administration was carried out in another study that used noxious thermal stimulation.

Placebo treatment affected opioid activity in several opioid-rich regions that play central roles in pain and affect, including the periaqueductal gray, dorsal raphe, nucleus cuneiformis, amygdala, orbitofrontal cortex, insula, rostral anterior cingulate, and lateral prefrontal cortex. Opioid activity in many of these regions was correlated with placebo effects in reported pain. Connectivity analyses of individual differences in opioid binding revealed that placebo treatment increased connectivity between the periaqueductal gray and the rostral anterior cingulate cortex and increased functional integration among limbic regions and the prefrontal cortex. Overall, the results suggest that endogenous opioid release in core affective brain regions is an integral part of the mechanism whereby expectations regulate affective and nociceptive circuits.

Our knowledge about the involvement of endogenous opioids in placebo analgesia has increased even more in the past few years. For example, it is now clear that a specific descending pain-modulating network mediates the analgesic effect following placebo administration. Eippert et al. combined naloxone administration with functional magnetic resonance (fMRI) imaging and found that naloxone reduced both behavioral and neural placebo effects as well as placebo induced responses in pain modulatory cortical structures such as the rostral anterior cingulate cortex. A brainstem-specific analysis also found a similar naloxone modulation of placebo induced responses in key structures of the descending pain control system, such as the hypothalamus, periaqueductal gray, and rostral ventromedial medulla. Interestingly, naloxone abolished the placebo-induced coupling between the rostral anterior cingulate cortex and the periaqueductal gray, which predicted both behavioral and neural placebo effects as well as the activation of the rostral ventromedial medulla. The same group found that the activation of this descending system following placebo administration extends to the dorsal horns of the spinal cord, although we do not know whether the spinal effects are mediated by endogenous opioids.

It has long been known that endogenous opioid systems are not the only mechanisms involved in placebo analgesia. One example of non-opioid-mediated placebo responses is represented by previous exposure to a non-opioid drug, such as ketorolac. When ketorolac is administered for two days in a row and then replaced with a placebo on the third day, the placebo analgesic response is not reversed by naloxone, indicating that specific pharmacologic mechanisms are involved in a learned placebo response, depending on the previous exposure to opioid or non-opioid substances. Other examples of placebo analgesic effect that are not mediated by opioids have been described in people with IBS and experimental pain in brain imaging studies. Based on these considerations, Benedetti et al. induced opioid or non-opioid placebo analgesic responses and assessed the effects of the CB1 cannabinoid receptor antagonist rimonabant. Unlike naloxone, rimonabant had no effect on opioid-induced placebo analgesia following morphine pre-conditioning, whereas it completely blocked placebo analgesia following non-opioid pre-conditioning with the nonsteroidal anti-inflammatory drug (NSAID) ketorolac. These findings indicate that the placebo analgesic responses elicited by NSAID conditioning are mediated by CB1 cannabinoid receptors.

Since the involvement of CB1 cannabinoid receptors in placebo analgesia is a recent finding, little is known about their localization and activation. To the best of our knowledge, they are activated following a previous exposure to NSAIDs, which suggests that these drugs, besides the inhibition of cyclooxygenase and prostaglandin synthesis, activate an endocannabinoid pathway. Indeed, non-opioid drugs, such as metamizole, have been found both to inhibit the enzyme cyclooxygenase and to act on the CB1 cannabinoid receptors in rodents. Based on these considerations, a change in the availability of endogenous ligands for CB1 cannabinoid receptors, such as anandamide, should be expected to modulate placebo analgesia in some circumstances. Indeed, Pecina et al. investigated the role of the common, functional missense variant Pro129Thr, of the gene encoding fatty acid amide hydrolase (FAAH), the major degrading enzyme of endocannabinoids, on psychophysical, dopaminergic, and opioid responses to pain and placebo induced analgesia in humans. FAAH Pro129/Pro129 homozygotes reported higher placebo analgesic responses and more positive affective states immediately and 24 hours after placebo administration. Pro129/Pro129 homozygotes also showed greater placebo induced µ opioid, but not D2/D3 dopaminergic enhancements in neurotransmission in regions known to be involved in placebo effects.

Neuroimaging of Placebo Analgesia

( Fig. 17. 1 )

Modern brain imaging techniques have been fundamental in the understanding of placebo analgesia, and many brain imaging studies have been carried out to describe the functional neuroanatomy of the placebo analgesic effect ^{, ,} As described above, the first imaging study of placebo analgesia showed that a subset of brain regions is similarly affected by either a placebo or a µ-opioid agonist. In an fMRI imaging study of experimentally induced pain in healthy participants, Wager et al. found that placebo analgesia was related to decreased neural activity in pain processing areas of the brain. Pain-related neural activity was reduced within the thalamus, anterior insular cortex, and anterior cingulate cortex during the placebo condition compared with the baseline condition. The magnitude of these decreases correlated with reductions in pain ratings. Not only did Wager et al. image the time period of pain as well as the time period of the anticipation of pain. They hypothesized an increase in neural activity within the brain areas involved in expectation. Indeed, they found significant positive correlations between increases in brain activity in the anticipatory period and decreases in pain and pain-related neural activity during stimulation within the placebo condition. The brain regions showing positive correlations during the anticipatory phase included the orbitofrontal cortex, dorsolateral prefrontal cortex, rostral anterior cingulate cortex, and midbrain periaqueductal gray. The dorsolateral prefrontal cortex is a region that has been associated with the representation and maintenance of information needed for cognitive control, consistent with a role in expectation. In contrast, the orbitofrontal cortex is associated with functioning in the evaluative and reward information relevant to the allocation of control, consistent with a role in affective or motivational responses to anticipation of pain. The anterior cingulate cortex is often reported to be involved in placebo analgesia, although some discordant results have been obtained. For example, it was found to have increased activity in a study by Petrovic et al. and decreased activity in a study by Wager et al. which might be explained based on the different experimental settings.

Most brain imaging studies aimed at investigating placebo analgesia have been performed in experimental settings in healthy volunteers. In contrast, Price et al. conducted an fMRI imaging study in which brain activity of IBS patients was measured in response to rectal distension using a balloon barostat. A large placebo effect was produced by suggestions and accompanied by large reductions in neural activity in the thalamus, primary and secondary somatosensory cortices, insula, and anterior cingulate cortex during the period of stimulation. It was accompanied by increases in neural activity in the rostral anterior cingulate cortex, bilateral amygdala, and periaqueductal gray. This study is important and informative, as it shows that placebos act on the brain in a clinically relevant condition in the same way as they do in the experimental setting. Therefore the involvement of key areas in placebo analgesia, such as the anterior cingulate cortex, is not limited to experimental noxious stimuli but also extends to clinical pain. The study by Price et al. is also interesting because reductions in brain activity occurred during the stimulus presentation itself, not just when subjects reported pain. It has been argued that the length of the painful stimulation may be critical for the measurement of placebo effects, as most studies used short heat or electric shock as pain stimuli and recorded activity decreases during periods extending after the stimulus offset. This may include a later cognitive reappraisal of the significance of pain and/or late neural activity influenced by report bias.

To determine whether the expectation of analgesia exerts its psychophysical effect through changes in the perceptual sensitivity of early cortical processes (in the primary and secondary somatosensory areas) or on later cortical elaborations, such as stimulus identification and response selection in the anterior cingulate cortex, Lorenz et al. used high temporal resolution techniques (magnetoencephalography). They found that activity in the secondary somatosensory cortex was highly correlated with the extent of influence of the subjective pain rating by pre-stimulus expectation. In contrast, anterior cingulate cortex activity seemed to be associated only with stimulus intensity and related attentional engagement. In another study on laser evoked potentials by Wager et al., early nociceptive components were found to be affected by placebos. Therefore a late cognitive reappraisal of the significance of pain and/or late neural activity influenced by report bias may not be responsible for this early modulation. This indicates that the early sensory components are affected by placebo manipulation.

Overall, all these brain imaging data were summarized using a meta-analysis approach with the activation likelihood estimation method. Nine fMRI studies and two positron emission tomography studies were selected for the analysis. During the expectation phase of analgesia, areas of activation were found in the left anterior cingulate, right precentral, and lateral prefrontal cortex, and the left periaqueductal gray. In the phase following pain stimulation, activation was found in the anterior cingulate and medial and lateral prefrontal cortices, in the left inferior parietal lobule and postcentral gyrus, anterior insula, thalamus, hypothalamus, periaqueductal gray, and pons. Conversely, deactivations were found in the left mid- and posterior cingulate cortex, superior temporal and precentral gyri, left anterior and right posterior insula, claustrum putamen, and right thalamus and caudate body. These meta-analytic data summarize all brain imaging studies and provide a global figure of the sequence of events following placebo administration. After the activation of a pain modulatory network during the expectation and early pain phases, several deactivations occur in different areas involved in pain processing. Further clarification to the functional neuroanatomy of placebo analgesia and the meta-analysis performed by Amanzio et al. comes from a more recent meta-analysis aimed at assessing whether the neurologic pain signature, which represent the set of brain regions involved in pain processing, is specifically affected by placebos. In this analysis, Medline (PubMed) was searched from inception to May 2015 and involved studies of functional neuroimaging of the human brain with evoked pain delivered under stimulus intensity-matched placebo and control conditions. Data were obtained from 20 of the 28 identified eligible studies, resulting in a total sample size of 603 healthy individuals. The responses to painful stimulation compared with baseline conditions were positive in 575 participants (95.4%). Placebo treatment resulted in significant behavioral outcomes on pain ratings in 17 of 20 studies (85%). However, placebo effects on the neurologic pain signature response were significant in only 3 of 20 studies (15%) and were very small. Similarly, analyses restricted to studies with low risk of bias indicated very small effects, and analyses of placebo responders indicated small effects. These findings indicate that placebo treatment has moderate analgesic effects on pain reports. The very small effects on neurologic pain signature are probably attributable to the fact that placebos affect pain via brain mechanisms largely independent of effects on bottom-up nociceptive processing. Placebos may act in regions other than the neurologic pain signature.

Brain connectivity has also been claimed to play a role in placebo analgesia, although this approach requires further clarification. For example, Tetreault et al. found that clinical placebo response is predictable from resting-state functional magnetic resonance imaging brain connectivity in patients with chronic knee osteoarthritis pain. However, whether brain imaging can predict who will respond to a placebo and who will not remain an open question.

The Reward Dopaminergic System

( Fig. 17.1 )

In a brain imaging study in which both positron emission tomography and fMRI imaging were used, Scott et al. tested the correlation between responsiveness to placebo and a monetary reward, using a model of experimental pain in healthy subjects. They found that placebo responsiveness was related to the activation of dopamine in the nucleus accumbens, a region involved in reward mechanisms, as assessed using in vivo receptor binding positron emission tomography with raclopride, a D2/D3 dopamine receptor agonist. The same subjects were then tested using fMRI imaging for monetary responses in the nucleus accumbens. These investigators found a correlation between the placebo responses and the monetary responses—the larger the nucleus accumbens responses to monetary reward, the stronger the nucleus accumbens responses to placebos. This study strongly suggests that placebo responsiveness depends on the functioning and efficiency of the reward system. This explains why some individuals respond to placebos, while others do not. Those who have a more efficient dopaminergic reward system would also be good placebo responders. Interestingly, Scott et al. used an experimental approach that is typical of clinical trials, whereby the subjects know they have a 50% chance of receiving either placebo or active treatment and whereby no prior conditioning was performed. Thus more powerful placebo responses should have been expected if the subject knew they had a 100% chance of receiving active treatment but actually received the placebo (deceptive administration) or if prior pharmacologic conditioning had been carried out.

In a different study by the same group, Scott et al. studied the endogenous opioid and dopaminergic systems in different brain regions, including those involved in reward and motivational behavior. The participants underwent a pain challenge in the absence and presence of a placebo with the expected analgesic properties. Positron emission tomography with 11C-labeled raclopride was used to analyze dopamine and 11C-carfentanil for analyzing opioids. Placebo induced the activation of opioid neurotransmission in the anterior cingulate, orbitofrontal and insular cortices, nucleus accumbens, amygdala, and periaqueductal gray matter. Dopaminergic activation was observed in the ventral basal ganglia, including the nucleus accumbens. Both dopaminergic and opioid activity were associated with both anticipation and perceived effectiveness of the placebo, as shown by the reduction in pain ratings. Large placebo responses were associated with greater dopamine and opioid activity in the NAc. Interestingly, nocebo responses were associated with the deactivation of dopamine and opioid release. The release of dopamine in the NAc accounted for 25% of the variance in the placebo analgesic effects. Therefore placebo and nocebo effects seem to be associated with opposite responses of dopamine and endogenous opioids in a distributed network of regions that form part of the reward and motivation circuit.