We all know that a patient’s positive expectation of treatment can sometimes create a positive result: That phenomenon is the basis for the placebo effect. Less appreciated is how a negative outlook can torpedo the efficacy of even a powerful analgesic. Several recent studies have explored the impact of positive and negative expectations, calling attention to the question: If expectations rule, what does that mean for clinical trials of new analgesics, or for optimal use of existing ones?
In a study that demonstrates the dumbfounding power of perceptions, Ulrike Bingel of University Medical Center Hamburg-Eppendorf in Germany and Irene Tracey at the University of Oxford in England showed that if subjects were conditioned to believe that an upcoming heat stimulus was going to be painful and were given the potent analgesic remifentanil without their knowledge, they reported feeling as much pain as if they had not been given the drug. Using whole-brain functional magnetic resonance imaging (fMRI), the investigators found that positive and negative expectations of relief engaged distinctly different areas of the brain. Their study appeared in the February 16 issue of Science Translational Medicine.
What was surprising is how the effects of a proven painkiller were obliterated by negative suggestion. “Remifentanil is a very powerful analgesic,” says Dennis Turk, an expert in clinical trial design at the University of Washington in Seattle. “This shows you the potency with which expectations can act.”
In the experiment, 22 healthy adult volunteers were monitored by fMRI while receiving a series of four painful heat stimuli. During the first run, the subjects received a saline infusion. Then, the saline was surreptitiously switched to remifentanil, and the subjects were given the same heat stimulus. In a third run, the subjects were again given the stimulus plus remifentanil, but this time, they were told they were getting the drug. In a fourth and final heat pulse, the subjects were falsely told that the remifentanil infusion had been stopped.
In this way, subjects underwent one baseline run, then their second, third, and fourth runs all involved an identical heat stimulus and a constant dose of remifentanil. What differed was what the subjects were told about whether they would be receiving the drug, and thus their expectation of relief. (In order to maximize expectations, the subjects were also conditioned beforehand: In an introductory session, the temperature had been secretly reduced in the third run, to bolster subjects’ positive expectancy. Then, the temperature had been ramped up in the fourth run and the remifentanyl switched to saline, to exacerbate negative expectations.) The effects of the different expectations were dramatic: When the subjects had positive expectations of analgesia, they reported less pain than in an expectation-free run. When the subjects had negative expectations of relief, their pain ratings were almost as high as the drug-free baseline.
Functional brain imaging studies over the last decade have revealed expectations at work as they sculpt a person’s perceptions of pain (for example, see the classic study of Wager et al., 2004). The effects involve not only the well-known “pain matrix,” but also a wide variety of other brain regions. Bingel, Tracey, and their colleagues saw that different brain regions became activated when patients had high hopes and enhanced analgesia (placebo), versus dread and dampened drug effects (nocebo). The placebo response was associated with activity in the endogenous pain modulatory system including the anterior cingulate cortex (ACC). Nocebo, on the other hand, was linked to activity in the hippocampus, an area that Tracey and her colleagues have previously linked to the exacerbation of pain by anxiety (Ploghaus et al., 2001). An earlier study also associated hippocampal activity with negative expectation and nocebo hyperalgesia in the context of sham acupuncture (Kong et al., 2008).
Placebo effects have generally been assessed by monitoring pain scores and brain activity when subjects are given inert compounds or sham therapies. This new study is notable because it measures placebo and nocebo responses to an active drug, say Randy Gollub and Jian Kong of Massachusetts General Hospital in Boston, both authors of the earlier study on sham acuptucture. “The study by Bingel et al. replicates the finding that negative expectancy with conditioning to elicit a nocebo response is associated with increased activity in the hippocampus and extends it in a fundamentally important way to the abolition of pharmacologically induced analgesia,” they write in an accompanying perspective
Considering clinical trials
The findings of Tracey, Bingel, and their team provide a cautionary tale about the potential impact of patients’ expectations on the success of pain therapy trials. Importantly, their study involved healthy participants, not patients with persistent pain. However, the researchers say that for patients with chronic disease—who may have had a long and frustrating history with failed treatments—negative expectations may fight against drug effects with even greater force. They write, “The underestimation of the influence that psychological states have on drug pharmacodynamics might… contribute to the frequent failure of clinical translation of drugs that show target engagement in preclinical studies, especially when drugs are developed for the treatment of chronic illness.”
The study also serves as reminder that in a treatment trial, the words of researchers can sway patients’ responses to the drug or therapy being tested. A particular problem is that different researchers can impart hugely different expectations, says Turk, who studies ways to sharpen trials through committees including the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT). “There’s been an assumption that we need bigger trials to get greater effects,” Turk told PRF. As a result, he noted, a single trial can pool results from a multitude of researchers around the world—each of whom may be saying something subtly different to their patients.
In a real-world example of the power of expectations, a study published online on February 20 in Pain by Nathalie André-Obadia and colleagues at INSERM in Lyon, France showed that subjects’ responses to a non-pharmacological therapy, repetitive transcranial magnetic stimulation (rTMS), can be heavily shaped by their previous experiences. The researchers found that a prior positive treatment experience could make a subsequent sham intervention just as effective as real therapy. If, on the other hand, the real treatment was not beneficial, then the following sham treatment actually worsened the patient’s pain.
Importantly, that study involved patients with chronic, treatment-resistant neuropathic pain, so it shows that patients who have suffered from prolonged pain remain heavily subject to placebo and nocebo responses. For clinical trials of new therapies aimed at relieving chronic pain, that means both positive and negative expectations are very much in play.
Comments
Megan Talkington, Pain Research Forum
As discussed in the article, the Bingel et al. study adds an important ingredient to placebo/nocebo studies: an active treatment. It measures the impact that expectations can have on subjects’ responses to a drug, not just to a sugar pill.
This points to a further question: How do placebo and active therapies interact?
Randomized controlled trials generally rest on the assumption that the two effects are additive. That way, the response in an experimental group, minus the response in a placebo group, indicates the contribution from the drug. It may not always be so simple, however. It could be that expectation and drug effects interact, such that the contribution from placebo is different for subjects receiving an active treatment. In a clinical trial, that would mean that the net effect of the drug could be over- or under-estimated.
A study by Predrag Petrovic, Karolinska Institute, Stockholm, Sweden, and colleagues, published last year in Pain (Petrovic et al., 2010), reported differences in regional brain activity elicited by placebo vs. opioid treatment. In assessing the results of that study, Tor Wager and Mathieu Roy of the University of Colorado at Boulder and Columbia University take it as evidence that placebo effects may be altered by active treatment. They write, “These findings are important and provocative because they provide evidence against the additivity assumption – in this case, the assumption that open-label remifentanil treatment produces an additive increment in effectiveness over placebo” (Wager and Roy, 2010).
At the recent NIH Pain Consortium Symposium, Lauren Atlas of the Wager group reported on a study seeking to address directly the question of how expectations and drug effects interact. Their poster presentation is attached here. Atlas and Wager found evidence that the two effects do not, in fact, interact. Instead, in their study, remifentanyl and expectancy contributed to analgesia additively, as judged by pain reports. Brain imaging revealed that remifentanil and expectancy separately affected pain-evoked responses in the ‘pain matrix’ regions of the brain, but their effects interacted in other regions. The results uphold the additivity assumption of placebo-controlled trials.