On April 26, 2017, PRF hosted its 25th webinar, which explored the science behind placebo and nocebo effects. Luana Colloca, University of Maryland, Baltimore, US, presented her talk, "From Learning Mechanisms to Expectancy Violation: Understanding Placebo and Nocebo Effects in Pain Research," to 150 live attendees from around the world. Placebo effects—the result of learning processes that depend heavily on expectation and psychosocial interactions—have the potential to boost existing pain treatments and give rise to new approaches in pain care.
The webinar was moderated by Tor Wager, University of Colorado, Boulder, US. Wager was joined by panelists Javeria Hashmi, Dalhousie University, Halifax, Canada, and Marta Pecina, University of Pittsburgh, US. (Listen to the webinar here).
Colloca began her talk by stating her goal of trying to understand the mechanisms underlying placebo and nocebo effects, with the ultimate objective to maximize benefits of placebo in many therapeutic interventions while minimizing nocebo effects. The term "placebo" refers to an inert substance used in testing new pharmacological agents to establish efficacy. Many drugs fail clinical trials because they don't outperform placebo interventions.
Colloca then clarified some terminology: “placebo response” is used in the context of clinical trials and refers to any symptom improvement seen in the group receiving a placebo treatment. Those improvements might be due to true “placebo effects”—the result of complex neurobiological changes—as well as any other factors, such as natural resolution of the condition. In the laboratory setting, researchers study placebo effects as changes resulting from patients’ perception, expectation, prior experience, and the therapeutic encounter around treatment. Most studies don’t include a “no-treatment group,” and don’t assess patient and provider expectations, but these steps are necessary to tease apart true placebo effects, Colloca said.
With any treatment—be it a drug, surgery, or other intervention—the psychosocial context is critical. It sets up an expectation of getting better (or worse, with nocebo), and it does so by activating the brain’s expectation pathways, Colloca said. She described a study she and her colleagues performed in which subjects undergoing surgery for lung cancer received an opioid drug either covertly—delivered automatically with no human interaction and so hidden from the patient—or overtly; in the latter case, information was provided by a healthcare worker whose presence created the psychosocial context (Colloca et al., 2004). Patients briefed about the drug showed faster and greater pain reduction after surgery. Moreover, when the drug was stopped, pain relief abated only when the patients were informed that they were no longer receiving it, demonstrating the power of expectation.
The chemical mediators of placebo
Many studies have investigated ways to understand and minimize nocebo effects, which suggest that healthcare providers should deliver truthful information about potential risks or negatives but manage expectations, tailor each encounter to the individual patient, and make education about placebo and endogenous pain-relieving systems part of the discussion. Studies have also long shown that endogenous opioids primarily underlie placebo effects, which are blocked by naloxone, a mu-opioid receptor antagonist. More recently, the neuropeptide hormones oxytocin and vasopressin have been implicated in placebo analgesia.
Colloca described another study using the tourniquet model of ischemic pain in which a blood pressure cuff is maximally inflated around the arm, which quickly becomes painful; most people can withstand the pain for about 10 minutes (Benedetti et al., 2007). After a dose of morphine, however, subjects were able to withstand the pain for more than 25 minutes. Given a placebo the following week, subjects’ tolerance for the pain fell somewhere between baseline and with opioid on board. Patients who received naloxone just before placebo, however, showed no placebo analgesia, indicating a role for endogenous opioid receptors in placebo effects.
In a more recent study of neuropeptide effects on placebo, Colloca and colleagues manipulated subjects’ expectations with verbal instructions (Colloca et al., 2016). Subjects were led to believe that an electric shock delivered to their arm would be more painful when a red screen appeared, but less painful when they saw a green screen (the shock was always of the same intensity). Subjects then rated their pain in response to the shocks, rating green-associated shocks as less painful than red ones, indicating a placebo effect. Next, they received an intranasal dose of vasopressin, oxytocin, saline, or no treatment. Forty minutes later, the researchers repeated the shock treatment. Only vasopressin powerfully enhanced placebo analgesia—that is, the tendency to feel less pain with green- compared to red-associated shocks—and only in women.
Learning mechanisms of placebo
Placebo and nocebo effects can be elicited in at least three main ways, Colloca explained: with verbal suggestion (instructional learning); social modeling (vicarious learning); and conditioning cues (experiential learning; see Colloca, 2014). Most importantly, focusing on expectancy when employing these techniques can lead to understanding the biological pathways responsible for the effects. “Indeed, we can use verbal suggestion, vicarious learning … and experiences to boost expectancy that dynamically updates our brain and our body to form placebo analgesic effects,” Colloca said.
Conditioning is the most effective inducer of placebo effects. For example, when subjects receive pain with a red light and no pain with a green light during conditioning, they later report lower pain associated with the green light even when both lights are paired with painful shocks in a test run (Colloca and Benedetti, 2006). Control subjects who did not undergo conditioning and were given no instructions rated green and red shocks the same. People given verbal manipulation alone (they were told that the green shocks would be less painful) rated pain during the green light as slightly less painful than did controls, but they did not show the robust placebo effect seen in those with conditioning. A similar experimental approach in a different study showed that nocebo effects, in contrast, were equally prevalent when subjects received verbal suggestion or underwent conditioning, indicating that different learning mechanisms may underlie the two phenomena (Colloca et al., 2008).
Observing another person receiving analgesia can set up a strong expectation of relief, leading to vicarious learning. In a 2009 study, Colloca again paired painful shocks with red lights and non-painful shocks with green lights, which subjects reported as such (Colloca and Benedetti, 2009). Observers of the procedure who then received painful shocks with both red and green lights also reported green-associated shocks as less painful than red-paired shocks, indicating vicarious learning that was comparable to direct experiential learning by those in the conditioning group.
To understand the brain mechanism responsible, Colloca then collected laser-evoked potentials from the skull, a measure of brain activity (Colloca et al., 2008). Placebo effects generated by either verbal suggestion or conditioning both resulted in changes in brain activity, even though verbal suggestion did not elicit large behavioral placebo responses. “That opens up many questions about the nature of consciousness and [its] interaction with the possibility to form placebo and analgesic effects,” Colloca said.
The learning mechanisms underlying placebo and nocebo may differ, but experiments with longer conditioning paradigms find that both effects persist over time. In a recent collaboration with Tor Wager, Colloca asked why placebo effects persist even after patients have been informed about the nature of those effects (Schafer et al., 2015). After one session of a conditioning paradigm, subjects displayed modest placebo analgesia, but after researchers revealed that they had received only a placebo cream and not a real analgesic, subsequent “treatment” with the cream produced nocebo (hyperalgesia). In subjects who underwent longer conditioning (four sessions), placebo analgesia was greater than in those with shorter conditioning, and after the reveal, subjects continued to benefit from placebo analgesia, suggesting distinct mechanisms underlying placebo from short versus long conditioning.
In another study, Colloca used functional magnetic resonance imaging (fMRI) while patients underwent conditioning and testing in the red/green screen paradigm (Lui et al., 2010). After conditioning, patients rated green-associated pain as lower than red-associated pain when they were at the same stimulus level, demonstrating that placebo analgesia had been established. Brain activity was modulated during green trials versus red trials in regions including the insula and the anterior and medial cingulate cortex. Most interestingly, participants with the greatest activation in the dorsolateral prefrontal cortex (DLPFC) during conditioning were those termed “placebo responders,” or people who demonstrated the greatest placebo effects.
Dashed expectations
What about the clinical reality in which patients’ expectations are often violated—that is, they receive something other than they thought, or expect something a clinician can’t deliver? Colloca modeled that situation with an unpublished study in which subjects were first conditioned with low-, medium-, and high-pain stimuli that were associated with a green, yellow, or red light and with a happy, neutral, or fearful face, respectively. In the test session, all painful stimuli were the same medium intensity.
As expected, the next day subjects responded with robust placebo and nocebo effects when the cue signals matched—i.e., when a green light and happy face appeared, subjects experienced stimuli as less painful; a red light and fearful face made stimuli more painful. But when the cues were mismatched—a green light with a fearful face, for example—placebo effects disappeared, and nocebo effects decreased significantly. Imaging data collected during the trial showed that key to these effects was activation in the left supramarginal gyrus, an area associated with detecting discrepancies in expectations.
Looking toward the future
Finally, Colloca turned to future work in the field and a recent Nature Outlook feature called “Placebos: honest fakery” by Jo Marchant, which examines the possibility of using the placebo effect to mitigate chronic pain (Marchant, 2016). “This is more than a dream,” Colloca said. She recently conducted a database search of studies using pharmacological conditioning with the aim of finding evidence of neurobiological mechanisms for placebo analgesia; she identified 10 studies in humans and animals (Colloca et al., 2016). In essence, pharmacological conditioning resembles classical Pavlovian conditioning, whereby a neutral stimulus such as a syringe is paired with a stimulus that produces a non-conditioned bodily reaction—pain relief following a drug—until the neutral stimulus itself produces analgesia.
How can this be applied in the clinic? Previous studies show that such conditioned responses have resulted in changes in the immune system. In pain medicine, conditioning with morphine might allow patients to take less of the drug. For example, a blister pack of pills containing a mix of active drug and placebo pills might provide relief even if patients are aware of the contents.
Highlights from the panel discussion
Is “placebo” a good word to use?
Panelist Javeria Hashmi wondered whether the term “placebo,” which many people conflate with something “ineffective,” will hinder how placebo effects can be used in the clinic. Colloca said that the benefit of using the current terminology is that it is familiar to people, although using terms like “expectancy-induced analgesia” may also be helpful.
Individual variation
Panelist Marta Pecina then commented that a great challenge in the field is to determine why some people are more susceptible to placebo effects than others, and why individuals vary in their placebo responses in different circumstances. Much research examines placebo responses to acute stressors, but the field needs to understand more about placebo in chronic conditions such as pain. Could answering the acute versus chronic question be a way to interrogate individual variability in placebo responses? Colloca replied that much work is underway to understand the transition from acute to chronic pain, including changes in brain circuitry, which will aid the understanding of placebo, and that different mechanisms probably underlie placebo in acute and chronic pain conditions.
From the lab to clinical trials
Colloca also said that researchers can apply what they have learned about placebo to design better clinical trials. First, Colloca said, they must measure expectations in providers and patients in trials. Second, they should use approaches such as brain imaging and modeling to understand and eventually predict who will respond to placebos. And finally, investigators need to better inform participants entering clinical trials about placebo effects in a way that minimizes those effects, so that drug efficacy can be properly assessed.
More on vasopressin
To another question from Pecina about the vasopressin work, Colloca replied that she and her colleagues have looked at the link between placebo and the hypothalamic-pituitary-adrenal (HPA) axis. They found that the women in the study who showed an enhanced placebo response with vasopressin also had reduced levels of cortisol, whereas participants with high cortisol and anxiety showed less placebo response with vasopressin, suggesting interactions with cortisol and stress. Vasopressin and oxytocin both play complex roles in sex-specific social behaviors, she added.
How long do placebo effects last?
Wager pointed to a question from an audience member: Can placebo effects last for a time long enough to be relevant to chronic pain even without continued conditioning or expectancy manipulation? He said the evidence shows that some forms of placebo can last very long. But the studies Colloca presented indicate that placebo effects are much like a learning process that should include extinction, but they don’t—why not?
Wager said that if a person expects less pain, and then experiences pain much worse than expected, the learning would be extinguished and placebo would go away. But if the “reinforcer” signal—the worse-than-expected pain—is actually diminished over time, extinction doesn’t occur, which is supported by some animal studies. So improvements in pain over time might effectively prevent extinction. Another possibility from psychology is the confirmation bias mechanism, which says that when people see positive effects that fit with their expectations, they will pay more attention to those aspects of the experience and less attention to those that conflict with their expectation, so they in turn perceive a greater effect. Shifting attention from negative to positive events in the clinic could help outcomes.
Placebo responders
Next, Wager framed a question for the panelists by noting that several companies are working to understand and predict which patients might be placebo responders—again, those who respond the most to placebo—and control for this in clinical trials. But are there actually placebo responders, and how might researchers identify them, considering some of the data Colloca mentioned about a proclivity to respond?
Colloca said that identifying responders is complex, but that her group has started to do detailed phenotyping, and to use imaging and genetic analysis to answer the question. The question should be framed in an integrative way, she said, where computational neuroscience and a search for biomarkers should be included. Hashmi commented that placebo responsiveness is probably a common developmental process, but whether a given individual has a response on a given day depends on many factors. Further, it does not make sense to exclude placebo responders from clinical trials; instead, they might benefit from lower doses or different treatments, she said.
Wager then asked a related audience question about placebo responders—is there a certain personality that makes people respond? Pecina said there seem to be some personality traits—being more agreeable, less hostile—that appear to make people more responsive to placebo, but these individuals also release more endogenous opioids following placebo administration, so there are biological factors at play, too. Hashmi added that placebo researchers are in the early stages of developing these ideas, and the brain mechanisms at play remain a black box. While computational neuroscience approaches will be important, researchers also need to keep in contact with clinicians to bring their laboratory findings into clinical practice, Hashmi said.
Placebo and integrative treatments
Another audience question asked about the relationship of placebo effects to psychotherapy, and to other integrative treatments such as acupuncture. Colloca noted findings from Pecina and colleagues showing that mu-opioids may be involved in placebo-mediated relief of major depression. With regard to acupuncture, placebo effects might be boosted because of attention to the patient and the context surrounding this treatment, which could improve analgesia. Every treatment, she added, has some placebo effect associated with it—that doesn’t mean it’s ineffective.
Wager also pointed to recent articles suggesting that the benefits of psychotherapy are due to placebo effects. By definition, he said, psychotherapy is a treatment, not an inert substance or sham treatment, so the benefits are not placebo effects. But the mechanisms of placebo and those that underlie psychotherapy, such as psychosocial interactions and rituals, overlap. Understanding which contextual processes are beneficial in pain conditions will aid application of knowledge from the lab to the clinic.
The ethics of placebo
What are the ethical issues about working with placebo effects and using placebos in place of active treatments? Colloca said, “From an ethical point of view, we need to be very careful when and how we use a placebo.” The American Medical Association has determined that it is unethical to deliver a placebo when an effective medication is available, but it is not necessary to use a placebo treatment per se in order to study the psychosocial context and other elements that invoke endogenous pain modulation, she said. Pecina added that even though placebos themselves shouldn't be used when a known effective treatment is available, there is a placebo effect in every intervention—there is no way to dissect the active drug effect from placebo or nocebo effects. Endogenous pain modulators, she said, are 20 to 30 times more powerful than morphine and without side effects—a fact that should help drive the research. Wager added that, rather than moving away from drug treatments entirely, researchers can engage placebo effects by adding psychosocial elements of treatment to augment the drugs.
Why are there placebo effects in the first place?
An audience member asked about the evolutionary purpose of placebo effects. Hashmi pointed to humankind’s evolutionary biology as hunter-gatherers. Traditional healers such as shamans used rituals and psychological elements as a crucial part of healing. “It’s a very strong part of the package in terms of how our brain works in relation to our disease, our longing for social contact when we are sick, and our need for other people to reach out to us when we are suffering,” she said; it's a mechanism to reduce isolation when people are most in need of being taken care of. Colloca added that placebo effects are really a powerful self-healing process, much like those in the body after injury. From an evolutionary point of view, this helps improve outcomes, and the social interaction element is critical.
Wager noted that Howard Fields' motivation decision model speaks to why there are placebo effects (see, e.g., Fields, 2004). Essentially, when thinking about pain, people need to decide which pain can be ignored and which pain is dangerous. If pain is thought of as a defensive response, then the pain individuals should feel really does depend on the broader context. From that perspective, placebo effects are adaptive, helping people choose the right response to pain.
Stephani Sutherland, PhD, is a neuroscientist, yogi, and freelance writer in Southern California.
Image credit: ayo88/123RF Stock Photo.