Putting placebos to the test

In the summer of 1919, a slight but otherwise healthy 15-year-old English Navy recruit succumbed to epileptic fits so severe, his doctor wrote, “he threw five strong men about like kittens.” Navy medics exhausted options to treat the boy, and sent him home. His family doctor prescribed a testicular extract on the theory that the boy’s seizures could be related to his delayed sexual development. The doctor also allowed that some pharmacologists dismissed such extracts “as having only a psychotherapeutic value as placebos,” though he thought the effect was real. (Sex hormones play a complicated role in adolescent seizures, for the record [1].) The pharmacologists likely embraced the definition of placebo offered a century earlier in Hooper’s Medical Dictionary as "any medicine adopted to please rather than to benefit the patient.” The curious power of the placebo received more scrutiny during the Second World War (Fig 1), when an American pharmacologist and anesthesiologist named Henry Beecher saw soldiers with shattered bones and shredded flesh rebuff his offers of morphine. A puzzled Beecher speculated that the men, so relieved to find themselves alive and safely removed from the horrors of battle, had entered a euphoric state that somehow overrode any feelings of pain, even after the ameliorative effects of shock had subsided. Since then, numerous studies have described placebo responses, mostly in neurological disorders, including Parkinson’s, depression, chronic pain, and epilepsy. As Beecher suggested, expectations can trigger the response, which as a 2007 PLOS Medicine study found, tends to be greater in children, who are more suggestible [2]. Still, the biological basis of the placebo effect remains obscure, leaving many to question whether reported clinical gains reflect genuine improvements or artefacts of statistics or physiology. Recent research in PLOS Biology [3] not only provides evidence that placebo treatment can produce significant pain relief but also flags a biomarker for the response. The placebo effect has been linked to neurobiological networks involved in expectations and inferences about what might happen based on past experience. So it stands to reason, the authors note, that the way these regions are wired in a person’s brain may influence their susceptibility to the response. To test this possibility, they scanned the brains of patients with chronic knee pain using functional magnetic resonance imaging (fMRI) before administering treatment in two small clinical trials. In the first, single-blind trial, patients were told they had a 50–50 chance of receiving a placebo or an active drug, though everyone received the placebo. The scans allowed the authors to identify brain differences between those patients who later responded to the placebo and those who did not. In the second trial, patients underwent brain scans before being randomly assigned to receive a placebo or duloxetine, a drug used to treat chronic pain. (In this randomized doubleblind trial, neither patients nor researchers knew who received which treatment.) The placeboresponse biomarker identified in the first trial again emerged in scans of placebo responders in the second. The finding of the same biomarker indicates that brain biology can indeed predict