Despite the enormous amount of time, effort, and money invested in testing new drugs, clinical trial failures are a common occurrence. According to a recent report a, the overall likelihood of approval for drugs in Phase 1 of development was less than 10%. For diseases with high placebo response (e.g., pain), failures are more common. For example, in 2011, clinicaltrials.gov listed 4,152 trials in progress for the treatment of pain, yet from 2008-2011, the only new drug approvals were given for new formulations or new dosage forms of existing drugs. b
Even getting a drug into Phase 3 is no guarantee of success, as such drugs only have around a 60% chance of moving on from there to the marketing application stage. a
Given this information, the question then becomes: how do drugs that appear to have very promising results from Phase 2, end up not confirming efficacy in Phase 3? A recent presentation at ASCPT discussed how increasing placebo response in Phase 3 may be at least partially to blame for Phase 3 failures.
Why Do Promising Drugs Fail in Phase 3?
In an effort to identify reasons for drug failures in Phase 3 despite encouraging Phase 2 results, the FDA summarized 22 cases where Phase 2 and Phase 3 trials had divergent results. These cases involved a wide range of medical products, including small molecule drugs, vaccines and other biologics, and devices. The FDA pointed out two main reasons for Phase 3 failures (among others):
- Use of biomarkers in Phase 2 that did not accurately predict the Phase 3 outcome (e.g., oncology and cardiovascular disease)
- Untested mechanism of action
While it is not mentioned as a potential reason in the FDA report, increased placebo response between Phase 2 and Phase 3 may be to blame for the failure to demonstrate efficacy in Phase 3 for some of these products. For example, four out of the 22 products in the FDA report (almost 20%) were being developed to treat diseases that have had a history of high placebo response. These products included Bitopertin as an add-on therapy in schizophrenia, a Capsaicin topical patch (Qutenza) for HIV induced nerve pain, Dexmecamylamine for depression, and Imiquimod (Aldara 5% cream) for molluscum contagiosum (a viral skin infection).
Confirmatory studies are double-blind and placebo-controlled because many patients may experience an improvement in symptoms while receiving a placebo treatment. While there is still uncertainty as to the exact reasons behind this phenomenon, patient expectancy is a key component. Indeed, once efficacy is achieved in Phase 2, both physician and patient expectancy increases, which can elevate the likelihood of a placebo response.
In addition to patient expectancy, a number of other factors may contribute to a placebo response. These include: the number of treatment arms (where the more likely the patient is to receive active treatment, the higher the placebo response), the number of therapeutic encounters, the patient’s genetics, and the duration of the clinical trial.
Of note, different indications can have different placebo response patterns. For example, high placebo response rates often occur in functional disorders (e.g., Irritable Bowel Syndrome) and those with imprecise endpoint measures (e.g., Crohn’s Disease), while placebo response tends to be lower for diseases like rheumatoid arthritis.
There is also evidence that the placebo response has steadily grown over the years. Authors from the FDA have summarized the placebo and treatment effects over time for successful drugs submitted in depression c and ADHD d, suggesting that drug effects also increase over time. However, the studies that failed to establish efficacy were not necessarily included in their assessment. Thus, in cases where response to active drug remains stable, increased placebo response in Phase 3 could result in an overall reduced treatment advantage for the investigational product.
Identifying the Causes of Placebo Response
While placebo response can be a significant contributor to clinical trial failures, the drug development community is still struggling to understand the precise mechanisms underlying this phenomenon. As for any complex problem, one of the most pressing concerns is to develop a better understanding of the root causes of the placebo response and to devise and implement strategies to address these causes.
To this end, an association of international scholars known as the Society for Interdisciplinary Placebo Studies (SIPS) has been established. The goals of this group are to use multidisciplinary tools (neuroscience, psychology, cognitive science, history, anthropology, and philosophy) to examine the physiological and psychological mechanisms underlying placebo effects and to develop ethically acceptable methods to harness the placebo effect to improve treatment outcomes.
Research from this group has demonstrated the effect of patient-physician encounters on placebo response, which was best described by Czerniak (2016) e, where 122 healthy volunteers were exposed to the cold pressor test from a doctor (actor) using two different scripts:
- Standard doctor-patient encounter (Treatment A), where the doctor acts busy and the volunteer has to wait, statements are “facts only”.
- Doctor is attentive and gives strong suggestions (Treatment B), where the doctor is empathetic and strongly suggests that this individual cream will work.
There was a 60% increase in the pain threshold with Treatment B. Thus, training investigators on ways to reduce placebo response through therapeutic encounters could reduce the placebo response and the potential for Phase 3 failures. This becomes even more important given that Phase 3 studies include sites around the world, many with varying degrees of access to health care.
In addition to the effect of patient-physician encounters, researchers from SIPS have identified an interaction between drug and placebo mechanisms. For example, a 2016 report f showed that brain imaging can be used to predict placebo responders and subjects with a higher propensity for response to active drug. The authors went on to show that placebo-response of brain connectivity accurately (95%) predicts placebo responders and nonresponders and showed that after active treatment, there were three subsets of subjects: (i) those where active drug enhanced predicted placebo response; (ii) those where active drug had no effect relative to predicted placebo response; and (iii) those where active drug diminished the predicted placebo response. Therefore, the implicit assumption made when evaluating the magnitude of a treatment effect (i.e., placebo response is the same in the drug arm and in the placebo arm and active-placebo response rate equals the true magnitude of drug effect) is violated given that there is an interaction between placebo and active drug.
The importance of genetic factors affecting placebo response is another important factor to be considered. Kathryn Hall of Brigham and Women’s Hospital and Harvard Medical School recently presented data suggesting that there are specific genes that may be responsible for placebo response, which are referred to as the placebome.
Using Study Design to Reduce Placebo Response
Thoughtful study design can be one way to reduce the placebo response. While several designs have been utilized in an attempt to reduce placebo response, the following study designs have generally proven ineffective for this purpose (i.e., they failed to improve the magnitude of efficacy):
- Including a run-in period, which is a period that occurs before the main part of the study and during which all patients receive placebo. The advantage of this design is that placebo responders and noncompliant patients can be eliminated, with the remaining patients being randomized 1:1 to placebo or active treatment.
- Employing a sequential parallel comparative design, in which the first half of the study has more than 50% of patients assigned to placebo, while in the second half, placebo non-responders are randomly assigned 1:1 to placebo or active treatments. The advantage of this design is again to weight the study in favor of placebo non-responders in an effort to reduce placebo response.
While intuitively these study designs might appear to be advantageous for reducing placebo response, they share a common flaw, which is that when placebo responders are eliminated, drug responders may also be eliminated.
Other study designs that have a greater potential to reduce placebo response include:
- Using a double-blind, variable duration design with a placebo run-in period. Using this design, both the patients and the clinical staff are blinded to the length of the placebo run-in period and the start of the active treatment period.
- Including a randomized withdrawal design, which involves randomly assigning patients who respond to the active drug either to continue with active drug or to be moved to placebo.
- Using hidden vs. open dosing. If the drug is effective, a reduction in symptoms should occur regardless of whether the patient is aware of treatment administration, with hidden dosing diminishing the impact of patient expectancy on treatment effect.
Using Alternative Methods of Analysis to Address Placebo Response
Another strategy that can be used to address placebo response is to employ alternative methods of analysis, including:
- Modeling placebo response and drug response
- Evaluating factors affecting placebo response
- Analyzing the outcomes of a trial using the center-specific level of placebo response as a weighting factor to estimate the treatment effect
- Controlling for factors that affect placebo response (e.g., baseline score, genetics)
Failures in Phase 3 clinical trials are costly and can be devastating for drug developers and for patients. An understanding of the root causes of the increased placebo response has the potential to reduce Phase 3 failures by identifying and mitigating placebo response risks. Among the options for reducing placebo response risk are to employ alternative study designs and methods of analysis.
Nuventra has a team of experienced drug development professionals and industry veterans who can provide critical insights into ways to reduce placebo response risk and give your clinical trial the best chance for success.
Contact one of our senior consultants today to discuss your study design.
a Thomas DW, Burns J, Audette J, Carroll A, Dow-Hygelund C, and Hay M. Clinical Development Success Rates 2006-2015. Biotechnology Innovation Organization, Biomedtracker, Amplion.
b Benedetti F, Carlino E, and Piedimonte A. Increasing uncertainty in CNS clinical trials: the role of placebo, nocebo, and Hawthorne effects. Lancet Neurology 2016;15:736-47.
c Khan A, Fahl Mar K, Khan Schilling S, and Brown WA. Has the rising placebo response impacted antidepressant clinical trial outcome? Data from the US Food and Drug Administration 1987-2013. World Psychiatry. 2017; 16(2):181-92.
d Khan A, Fahl Mar K, and Brown WA. Does the increasing placebo response impact outcomes of adult and pediatric ADHD clinical trials? Data from the US Food and Drug Administration 2000-2009. J Psychiatr Res. 2017; 94:202-7.
e Czerniak E, Biegon A, Ziv A, Karnieli-Miller O, Weiser M, Alon U, and Citron A. Manipulating the Placebo Response in Experimental Pain by Altering Doctor’s Performance Style. Front Psychol. 2016; 7:874.
f Tétreault P, Mansour A, Vachon-Presseu E, Schnitzer TJ, Apkarian AV, and Baliki MN. Brain Connectivity Predicts Placebo Response across Chronic Pain Clinical Trials. PLoS Biol. 2016; 14(10):e1002570.