Biology: Der Placebo-Effekt

A meta-analysis of clinical-trial data submitted to the US regulator (including unpublished trials) examined how the antidepressant-versus-placebo difference varied with baseline severity. Both drug and placebo groups improved substantially; the difference between them met a standard threshold for clinical significance only at the highest severity levels, and this widening gap was driven mostly by a diminishing placebo response in severe depression, not a rising drug response. The finding measures how large the placebo component of the response is, not that antidepressants are ineffective.

Two complementary lines establish the mechanism. In post-surgical dental pain, administering naloxone (an opioid receptor antagonist) abolished much of the analgesia produced by placebo, implicating endogenous opioids (Levine 1978). Decades later, functional MRI combined with naloxone showed that placebo analgesia activates the descending opioidergic pain-modulatory system, from the rostral anterior cingulate cortex through the periaqueductal gray to the rostral ventromedial medulla, and that naloxone reduces both the behavioral and the neural placebo response (Eippert 2009). Together they show placebo pain relief is a real, opioid-dependent neurobiological process.

A systematic review and meta-analysis reviewed 13 randomized controlled trials of open-label placebo (openly described inert treatment) and pooled 11 of them across a spread of conditions and found a significant positive effect overall in favor of open-label placebo compared with no treatment. Effects were clearest for self-reported, brain-mediated symptoms. The authors note the trials are mostly small, short, and heterogeneous, so the pooled estimate establishes that the effect is real and reproducible without fixing a precise size.

Using positron emission tomography with raclopride to index dopamine release, researchers found that placebo administration in patients with Parkinson's disease produced marked release of endogenous dopamine in the dorsal and ventral striatum. The response was linked to the patients' expectation of therapeutic benefit and to activity in reward circuitry. It is a small mechanistic imaging study, and it provided direct biological evidence that a placebo response engages a specific, disease-relevant neurochemical system.

Two major reviews consolidate the field. A Lancet review describes placebo effects as psychobiological events attributable to the therapeutic context, produced by expectation and by classical conditioning, and cautions that they are not a nuisance to be dismissed (Finniss 2010). A Nature Reviews Neuroscience review maps the neuroscience: how context and learning shape health outcomes, the opioid and dopamine systems involved, individual variation, and the nocebo counterpart (Wager 2015). Both frame placebo as several mechanistically distinct processes, not one uniform effect.

Reviews of the placebo literature document the nocebo counterpart: verbal suggestion of a negative outcome, or prior negative experience, produces symptom worsening, including increased pain (hyperalgesia), nausea and fatigue, with measurable changes in cholecystokinin, dopaminergic and other pathways. A practical consequence is that a large fraction of the adverse effects reported by patients on active medication also appear in the placebo arms of trials, indicating an expectation-driven component. The nocebo effect shows the mechanism is bidirectional.

The SAMSON trial gave 60 participants who had abandoned statins twelve one-month bottles: four of atorvastatin 20 mg, four of placebo, and four empty, in random order, with daily symptom intensity logged in an app. Forty-nine completed the 12-month protocol. Symptom scores were much higher than baseline both on the statin (16.3) and on the placebo (15.4) than in no-tablet months (8.0), and the statin and placebo did not differ (P=0.388), giving a nocebo ratio of 0.90. Neither the intensity of symptoms on starting nor their relief on stopping distinguished statin from placebo. Six months after the trial, half the participants had restarted statins. It shows most of the symptoms attributed to a real drug can be nocebo, driven by expectation, not the chemical.

A randomized controlled trial in 80 patients with irritable bowel syndrome compared an open-label placebo, presented openly as inert pills with an explanation that placebos can produce powerful mind-body self-healing effects, against a no-treatment control receiving the same clinician attention. After three weeks the open-label placebo group scored significantly higher on the IBS Global Improvement Scale (5.0 vs 3.9, p=0.002) and were more likely to report adequate symptom relief (59% vs 35%). It is a small, short trial in a symptom that reports subjectively, and it is the foundational demonstration that placebo effects do not require deception.

A randomized controlled trial in 97 patients with chronic low back pain compared three weeks of open-label placebo added to treatment as usual against treatment as usual alone. The open-label placebo, presented openly as inert, produced significant reductions in composite pain (about 30% in both usual and maximum pain) and in Roland-Morris disability, compared with little change in the usual-care arm. A follow-up in which the control group later crossed over to open-label placebo saw similar improvements. It is a small, short trial with self-reported outcomes.

A randomized controlled trial in 74 cancer survivors with persistent fatigue compared three weeks of open-label placebo against treatment as usual. The placebo group reported significantly greater improvement in fatigue severity (about 29%) and in fatigue-disrupted quality of life. Control participants who then took open-label placebo for three weeks showed comparable gains. The trial is small and short, and fatigue is a subjective outcome, but it extends the open-label placebo finding to a distressing symptom in a serious illness.

In a controlled experiment, healthy men received the immunosuppressant drug ciclosporin A together with a novel-tasting drink over an initial learning phase. On later re-exposure to the drink alone, without the drug, they showed suppressed immune function, including reduced interleukin-2 and interferon-gamma mRNA expression and reduced lymphocyte proliferation, compared with controls. This demonstrates that a placebo response can be conditioned into the immune system, one of the two core mechanisms (alongside expectation) behind placebo effects.