In drug development, placebo-controlled trials are the gold standard as they provide the strongest possible evidence that efficacy of the experimental therapy is causally related to the therapy itself. The use of placebo controls allows drug developers to account for the natural course of the disease over the trial period and the placebo response – the clinical improvement seen in patients after being given a sham or dummy treatment.
Historically, interpretation of clinical trials relies on “assay sensitivity”, or the sensitivity to detect clinically meaningful differences between endpoints measured in the group of patients given active drug compared to the group of patients given placebo. Assay sensitivity can be influenced by many factors, including the study design, specific endpoints selected, number of clinical sites and, of course, the magnitude of the placebo response. Trial statisticians conducting power analyses and sample size estimates while designing a trial typically assume an expected placebo response rate based on historical experience or literature values. A larger than expected placebo response can compromise assay sensitivity and lead to a failed or inconclusive trial, which may ultimately mean that the trial must be repeated or the program abandoned. In this context, the magnitude of the placebo response and its impact on assay sensitivity is viewed as a phenomenon of the placebo group, or the aggregate result of all patients receiving placebo in the trial.
Trial failure due to unexpectedly large placebo response rates is still all too common – including phase 3 trial failures1– despite decades of research into the placebo response and a variety of strategies (ranging from complex trial designs and training of clinical sites to avoid expectation inflation) aiming to mitigate the placebo response. Perhaps its time to start thinking about the placebo response slightly differently, and shift focus from the response of the placebo group to the placebo response of the individual patient.
Each patient in a clinical trial (including those receiving active drug) will have a placebo response that is unique to them and that is influenced by that patient’s individual characteristics. Hundreds of reports in the literature highlight the contribution of individual patient characteristics – such as age2,3, gender2, expectation for improvement4–7 and personality traits8–11. In reality, trials or indications with high placebo response rates in the placebo group result from both a high number of placebo responders and the magnitude of those patients’ placebo response.
In patients receiving experimental therapies, some component of the measured treatment response can likely be attributed to that patient’s unique placebo response. This has historically been difficult or impossible to measure. The range of placebo response among all patients in the clinical trial population not only contributes to the overall placebo response rate of that trial, it is also a main driver of data variability in both the placebo-treated group and the drug-treated group.
As the placebo response is actually related to each individual patient – and is not a characteristic of only the placebo-treated group – would it be feasible to consider each patient’s innate placebo responsiveness as a baseline characteristic for which adjusted analyses could be performed? For certain, statisticians routinely account for other innate patient characteristics when analyzing clinical trial data. For example, body weight and/or BMI may be used as a covariate in analysis of clinical trial data in schizophrenia 12 to reduce variability in treatment response seen in patients with varying BMI. Similarly, if it could be assessed at basline, each patient’s placebo responsiveness should be considered in clinical data analyses. Placebell©™ is indeed a method that can be used to predict placebo responsiveness based on individual patient characteristics and, as such, could be used to conduct adjusted analyses of clinical trial data. This approach uses machine learning to combine baseline data – including patient psychological traits and expectations, demographics (age, gender), medical history, baseline disease intensity, among other inputs – into one composite score reflecting each patient’s placebo responsiveness. Using the Placebell©™ score as a covariate can reduce data variance and increase trial power and, over time, can be used to reduce sample size. This would indeed improve clinical trial assay sensitivity and could be an effective measure to reduce the risk of trial failure related to the placebo response.
References
1. Dumitrescu TP, McCune J, Schmith V. Is Placebo Response Responsible for Many Phase III Failures? Clinical Pharmacology and Therapeutics. 2019;106(6):1151-1154. doi:10.1002/cpt.1632
2. Weimer K, Colloca L, Enck P. Age and sex as moderators of the placebo response – An evaluation of systematic reviews and meta-analyses across medicine. Gerontology. 2015;61(2):97-108. doi:10.1159/000365248
3. Ho TW, Fan X, Rodgers A, Lines CR, Winner P, Shapiro RE. Age effects on placebo response rates in clinical trials of acute agents for migraine: Pooled analysis of rizatriptan trials in adults. Cephalalgia. 2009;29(7):711-718. doi:10.1111/j.1468-2982.2008.01788.x
4. Pollo A, Amanzio M, Arslanian A, Casadio C, Maggi G, Benedetti F. Response expectancies in placebo analgesia and their clinical relevance. Pain. 2001;93(1):77-84. doi:10.1016/S0304-3959(01)00296-2
5. Wei H, Zhou L, Zhang H, Chen J, Lu X, Hu L. The Influence of Expectation on Nondeceptive Placebo and Nocebo Effects. Pain research & management. 2018;2018:8459429. doi:10.1155/2018/8459429
6. Darragh M, Booth RJ, Koschwanez HE, Sollers J, Broadbent E. Expectation and the placebo effect in inflammatory skin reactions. A randomised-controlled trial. Journal of Psychosomatic Research. 2013;74(5):439-443. doi:10.1016/j.jpsychores.2012.12.010
7. de Pascalis V, Chiaradia C, Carotenuto E. The contribution of suggestibility and expectation to placebo analgesia phenomenon in an experimental setting. In: Pain. Vol 96. Pain; 2002:393-402. doi:10.1016/S0304-3959(01)00485-7
8. Peciña M, Azhar H, Love TM, et al. Personality trait predictors of placebo analgesia and neurobiological correlates. Neuropsychopharmacology : official publication of the American College of Neuropsychopharmacology. 2013;38(4):639-646. doi:10.1038/npp.2012.227
9. Zhou L, Wei H, Zhang H, et al. The Influence of Expectancy Level and Personal Characteristics on Placebo Effects: Psychological Underpinnings. Frontiers in Psychiatry. 2019;10(FEB):20. doi:10.3389/fpsyt.2019.00020
10. Geers AL, Helfer SG, Kosbab K, Weiland PE, Landry SJ. Reconsidering the role of personality in placebo effects: Dispositional optimism, situational expectations, and the placebo response. Journal of Psychosomatic Research. 2005;58(2):121-127. doi:10.1016/j.jpsychores.2004.08.011
11. Colloca L, Klinger R, Flor H, Bingel U. Placebo analgesia: Psychological and neurobiological mechanisms. Pain. 2013;154(4):511-514. doi:10.1016/j.pain.2013.02.002
12. Deberdt W, Lipkovich I, Heinloth AN, et al. Double-blind, randomized trial comparing efficacy and safety of continuing olanzapine versus switching to quetiapine in overweight or obese patients with schizophrenia or schizoaffective disorder. Therapeutics and Clinical Risk Management. 2008;4(4):713-720. doi:10.2147/tcrm.s3153