So the last little while, I’ve been doing some thinking on why people, with seemingly homogeneous conditions, get such different responses to treatment. Specifically, I’ve been thinking about the role of patient expectation. To facilitate a chat on this topic, I’d like to discuss a paper that was recently published by the Oxford fMRI gang that evaluated the effect of treatment expectation on drug efficacy.1 This paper, typical of the work of Irene Tracey, used an elegant design to test the effect of patient expectation on reported pain levels when a potent opioid, Remifentanil, was provided. Now a study that just gave a drug and measured pain levels would be quite boring and not that informative. So instead, these researchers deceived the patients.
How did they do this? First, they infused the participants with normal saline and took pain ratings when noxious heat was provided (#1). Then, unbeknownst to the participant, the researchers started the remifentanil infusion. Thus when the researchers asked for pain ratings again (#2), they were assessing only the biological effect of remifentanil (ie, absent of treatment expectation). Then in the third run, they told the participants that they were starting the analgesic and that the participants should expect pain relief (ie, inducing positive expectation) and took a pain rating (#3). Finally, in the fourth run, they told the participants that they were stopping the analgesic and that the participants should expect to feel an increase in pain (ie, inducing negative expectation) and then took a pain rating (#4). Except in this last run, they didn’t stop the infusion of remifentanil…
What they found was fascinating. Pain ratings were decreased in #2 (didn’t know the drug started) compared with #1 (saline). This suggests that there is a biological mechanism of action of the drug (that reduces pain) that is independent of treatment expectation. Second, pain ratings were decreased in #3 (ie, positive expectation – told the drug would start and would decrease pain) compared with #2. This suggests that positive expectation provides additional treatment efficacy on top of biological mechanisms. Third, pain ratings were increased in #4 (ie, negative expectation – told the drug would stop, but it actually didn’t) compared with #3. This suggests that negative expectation can decrease the effectiveness of the drug. But this wasn’t even the most intriguing finding… Not only did negative expectation increase pain levels, but it completely negated the biological effect of remifentanil, returning pain levels to that similar to the saline pain ratings (saline pain ratings: 66 +/-2; negative expectation pain ratings: 64 +/-3). What????
First, we have to be a bit cautious because the order in which the conditions were given was not randomised, meaning that negative expectancy was always assessed last, where potentially, pain levels could increase due to opioid tolerance. However, the researchers also performed control experiments on the natural time course of the drug as well as sensitisation and habituation control experiments, which suggested that the findings were not likely to be due to opioid tolerance or habituation to remifentanil. Second, we are talking about healthy participants, in which experimental pain was induced using heat. The relevance of this to a clinical situation thus can be a bit of a stretch. Third, the model of negative expectation (ie, removing pain medication and expecting pain levels to increase) may not be directly relevant to clinical negative expectation (ie, getting this treatment will increase my pain levels). In other words, the negative expectations toward getting a treatment (clinically) versus having one taken away (as was performed in the present study) may not be the same thing.
However, despite these limitations, I think this paper raises some fascinating issues. For me it is particularly challenging because we are talking about reversing the effect of a potent opioid only by convincing people that you stopped the drug. It also makes me wonder – what is the neurobiology behind this? Is it a function of anxiety (ie, as in the present study anxiety levels were reduced during positive expectancy and increased during negative expectancy) and cholecystokinin-mediated? The authors also performed brain imaging on all subjects and there were differences in brain activation between the experimental conditions, but what does this mean? I’ve got my fingers crossed that hopefully the next studies evaluate the biology behind this effect as well as its translation into a clinical situation. Or maybe I’ll just take away the incredibly distorted message that pessimists should never do drugs.
1. Bingel U, Wanigasekera V, Wiech K, Mhuircheartaigh RN, Lee MC, Ploner M, Tracey I. The effect of treatment expectation on drug efficacy: Imaging the analgesic benefit of the opioid Remifentanil. Science Translational Medicine 2011; 6: 70ra14. DOI: 10.1126/scitranslmed.3001244