Historically, clinical trials have stuck rigidly to formal phases. Phase I trials test new drugs or devices with small groups of people. Phase II trials study larger populations and implement placebo groups. Phase III trials study even larger populations and can be used to seek FDA approval. Phase IV follows approval and to establish the long-term effects of the drug or medical device use.
Of course, these distinct phases remain important for study designs and protocols but the boundaries are blurring. Researchers are increasingly relying on adaptive design, which simply affords greater flexibility. After all, it allows researchers to use data collected during a study to inform changes in a trial’s design and analysis while the trial is still in progress.
We explore what seamless trials look like and how adaptive designs can enhance them by lowering costs and risks and improving results. But we also assess other study choices that can reduce the size of the control groups, drop costs and drive up efficiency, relaxing the strict divisions between trial phases.
A challenge with phase II trials, for example, is that only patients with the condition or illness being investigated can participate. This means it’s harder to find patients, writes Becky Simon at productivity platform SmartSheet. So instead of the typical 300 participants found in most phase II studies, there will usually be less than 100 in phase II cancer trials.
Simon says a proposed industry solution is to standardize the division of phase II trials in to “phase IIa” and “phase IIb trials.” The focus has been on HIV vaccine and drug trials in particular. Dividing these trials would require IIa to incorporate all of the normal activities of a phase II trial, while IIb would target dose-efficacy.
The separation of these goals can help save sponsors money if the drug does not work out, Simon says. She notes that phase II trials tend to be when new drugs most often fail yet they can cost around $20 million to run.
The old model of research does not work anymore. It’s costly and time-consuming with a low success rate for the investigational treatment. What is needed is an innovative approach from clinical trial sponsors and CROs, according to research by Katarina B. Pavlović, Ivana Berić and Ljiljana Berezljev published in the European Project Management Journal.
Their solution is to learn from software development and implement agile transformation into clinical research. Agile methodologies in clinical trials would require trial managers to think of medicine as software, which evolves as end-users interact with the product. As the evolution occurs, designs can be tinkered with.
So applied to adaptive designs, this means amending design or analysis based on data accumulated in the trial — researchers might reduce patient numbers or trial duration, for example.
The traditional model of parallel-group randomized controlled trials has served the clinical research community well over the years. But new types of research questions that focus on comparing multiple interventions or the effects of a therapy across different subgroups need a different approach, say researchers at Nature.
Other new clinical research goals include looking at related conditions or reducing downtime between trials, they explain. This is why adaptive platform trials, which break down the distinct barriers between trials phases, can extend the reach of the research. The researchers explain that APTs enable the study of multiple interventions in a disease on an ongoing basis, discarding or including therapies based on a predefined decision algorithm.
Doctors Brian M. Alexander and Lorenzo Trippa explain how such an algorithm works. A study would use multiple experimental arms categorized according to a genomic biomarker. After each arm is equally randomized, further randomization probabilities change as Bayesian-estimated results of the biomarker-specific probability of treatment are gathered. Arms could be added or removed as probability of treatment success is determined.
While many in the industry regard adaptive trial design as an important means of improving cost-efficiency and patient treatment, the viewpoint is far from unanimous. Indeed, Dr. Gail A. Van Norman says researchers must consider the ethics of adaptive trials.
Randomization, while considered a means of achieving valid results, is not necessarily ethical. For instance, a researcher would need to know that both arms of a study are effective if they were to assign a patient in need of treatment to either one of them. You can’t assign a patient to a treatment arm that won’t be effective, she argues.
Adaptive design appeals to healthcare statistician Marvin Zelen’s “play the winner” rule, Van Norman adds. This basically says that if a patient responds favorably to treatment in a comparative study, the next patient should be allocated to that study arm. The result would be more patients being assigned to the successful arm, fewer patients receiving ineffective treatment, and a shorter trial.
Yet, as Van Norman explains, this may not make adaptive trials more ethical. For instance, she says patient burden is not removed as the intermediate markers adaptive trials rely upon are only validated towards the end of phase II or III trials. Patient burden is only reduced if adaptive design means that phase II trials lead to faster approval of drugs. And she says the results are out as to whether or not this is the case.
Small biotech companies could benefit the most from adaptive designs in their trials but it’s the large pharma companies seizing the approach. And this needs to change, according to independent drug development consultant Dr. Scott Harris. The reason the larger companies are taking this approach is that they have more resources such as statisticians familiar with adaptive design and are better able to bear the risk involved.
The culture at these smaller organizations also inhibits them from seizing the opportunity, says Harris. “Within a small company, very often companies transition from scientific founders or scientific concepts at universities or somebody who has invented a concept in an independent laboratory. Transition from science to actual implementation is much more of a cultural challenge for a biotech than it would be for a big pharma company.”
While adaptive trials tend to test one drug on multiple patient subgroups, multi-arm trials compare multiple therapies against a common control group in one study. Such an approach is useful as it requires fewer patients, write Thomas Jaki and James M.S. Wason at BMC Cardiovascular Disorders.
Further, the result of this method is better comparison of drugs within the same study. This is due to using the same patient population, comparison group and study protocol.
A trial with three experimental arms rather than three separate two-arm trials results in 15 percent fewer patients in the control group, explain the researchers. This approach leads to more patients being randomized for experimental treatments, which alone will help in patient recruitment, a perennially difficult task.
Pursuing personalized medicine also cuts through traditional phase divisions of trials. Instead the effects of a drug on large populations, researchers target specific groups of people with similar genetic mutations.
Another goal of personalized medicine is to diagnose subtypes of disease based on genetic sequencing, explains David Goldstein, director of the Institute for Genomic Medicine at Columbia University Medical Center in New York. The result is fewer patients are needed per trial and the treatment is far more targeted. Researchers are able to determine whether a specific drug works on multiple subgroups based on genetics.
Researchers can avoid hiring a control group for trials, saving both money and time, by using real-world evidence. Jennifer Goldsack, interim executive director at the Digital Medicine Society, calls this approach a synthetic control arm.
Instead of researchers busying themselves collecting data from recruited patients in a control or standard-of-care group, they use the RWE — collected from electronic health records, historical trial data, administrative claims data, patient-generated data from wearables and disease registries — to create a comparator group.
Goldsack says this hybrid trial design will benefit pharma from improved efficiency and fewer delays. Plus there’s also a key advantage for patients. By using RWE instead of a control group, no patient will receive a placebo. This is important as fear of being given a placebo treatment deters people from participating in trials. “The use of synthetic control arms can also eliminate the risk of unblinding when patients lean on their disease support social networks, posting details of their treatment, progress, and side effects that could harm the integrity of the trial,” she adds.
Images by: kasto/©123RF Stock Photo, wrightstudio/©123RF Stock Photo, leaf/©123RF Stock Photo