Behind the curtain: the key role of biostatistics in advancing clinical trials

The journey from bench to bedside relies on clinical trials to test the safety, feasibility, and efficacy of medical advancements. These trials involve extensive collaboration to leverage interdisciplinary expertise, ensuring that whatever product reaches patients will be safe and beneficial. Biostatistics, an indispensable discipline positioned to ensure the integrity and validity of clinical trial outcomes, is an integral part of this collaboration. 

To be scientifically sound, a clinical trial must be designed from the very start according to specific parameters in line with pre-specified aims. Biostatisticians help guide the way, making sure that clinical trials collect the most robust data while prioritizing patient care. Biostatisticians work 'behind the curtain' as the unseen choreographers, ensuring that the steps in the dance of medical breakthroughs are choreographed to transform scientific inquiry into real-world advancements.

"People often think that biostatisticians only analyze data once the trial is complete. We certainly do that, but our role is much broader and starts much earlier," said Arzu Onar-Thomas, PhD, St. Jude Department of Biostatistics. "Biostatisticians are a central part of the study team and help formulate the questions the study is trying to answer." Onar-Thomas oversees the biostatistics support for all brain tumor treatment trials designed and run at St. Jude. She also serves as the Lead Statistician within the Children's Oncology Group (COG) Central Nervous System Committee and leads the Operations, Biostatistics and Data Management Core of the Pediatric Brain Tumor Consortium, whose primary objective is to rapidly conduct novel clinical trial evaluations of new therapeutic agents and technologies in children with primary central nervous system tumors.

Biostatistics start clinical trials off strong

Biostatistics is a branch of mathematics that applies statistical methods to topics and develops new methodologies applied to these areas. It encompasses designing experiments, analyzing data, and ensuring reliability in clinical trials by defining outcome measures, estimating appropriate sample sizes, improving randomization and minimizing bias. In clinical trials, these aspects are incorporated into the initial design and are integrated into the trial implementation.  

"Biostatisticians prospectively define a statistical analysis plan to evaluate trial outcomes for treatment efficacy. The prospective nature of the design and the analysis plan makes the trial results more reliable compared to other types of studies," Onar-Thomas explained. 

A well-planned and executed clinical trial is not immune to challenges. St. Jude has historically focused on rare pediatric diseases, and issues such as sample size and patient heterogeneity can pose significant challenges to trial design.  

Biostatisticians can address these issues by incorporating predefined adaptations into the trial design. In the simplest case, adaptive design establishes predefined time points to review the data, known as interim analyses, and determine whether or not to continue the trial as designed based on specific thresholds.  

"Adaptations have been commonly used since the early days of clinical trials. What's different now is that we can incorporate more extensive adaptations thanks to the computational power we have access to. These can lead to more nimble studies, though they also tend to be more complicated," Onar-Thomas said. "Planned adaptations allow early looks at the data and establish decision criteria based on early results to ensure the trial remains ethical and is on track to achieve its stated goals. They also allow the incorporation of new information that becomes available during the trial."

"There are always compromises involved with the investigation of rare diseases. A controlled randomized trial is the gold standard, but sometimes it is simply not feasible," Onar-Thomas explained. "Sometimes, a trial design is primarily guided by the limitations in sample size and may be less optimal statistically. In those cases, we prioritize the chance of appropriately answering the key trial questions with the understanding that we may be less confident in the rest of the results." 

Innovative designs for improved outcomes

With increasing complexity in clinical trial design – it is important to continue innovating and building new statistical tools to keep up with demands. Clinical trials are conducted in phases: phase I tests safety, phase II tests preliminary efficacy, and phase III tests superiority in a confirmatory manner. Haitao Pan, PhD, Department of Biostatistics, emphasizes the importance of biostatistics to making sure trials advance through the phases. 

Establishing optimal doses of new drugs, including the maximum tolerated dose (MTD) and the optimal biological dose (OBD), to maximize effectiveness and minimize side effects is an important part of early-phase clinical trial design. Pan developed user-friendly software tools and packages to design efficient dose-finding phase I and phase II clinical trials. Examples include Keyboard, an R package or toolkit for the R programming software, offering tools for identifying both the MTD and OBD of a new drug or combination of treatments, and UnifiedDoseFinding, an R package that extends conventional dose-finding trials to include non-binary toxicity endpoints. Additionally, an R package, frequentistBinary, and methods specifically designed for non-controlled multi-arm trials, particularly useful in pediatric oncology, were recently utilized in the St. Jude RMS2021 study protocol for rhabdomyosarcoma.

"We must intelligently utilize data to better assign incoming patients the appropriate dosages. This is especially important in pediatric oncology where even phase II studies can take up to 10 years because there are very few patients," Pan said. "We streamlined the process by integrating phase I data to inform phase II to enhance our understanding of the toxicity and efficacy of the treatment, ensuring we don't overlook valuable insights as we progress to later phases."

Pan is also developing novel approaches to later-phase clinical trial design. His designs make use of existing data and account for the unique features of immunotherapy, allowing for more flexibility when conducting dose-finding studies and testing new treatments. For example, one methodology, published in the New England Journal of Statistics in Data Science, allows researchers to incorporate new treatment arms or participant groups into ongoing trials without compromising acceptable error rates. This accommodates the implementation of newly available treatment options without having to arrange a separate trial, potentially expediting the approval of effective treatments. 

Pan’s team has developed and released a total of 13 R packages for clinical trial design in recent years to meet St. Jude's needs. These advancements in clinical trial design are further elaborated in his new book, "Bayesian Adaptive Design for Immunotherapy and Targeted Therapy."

Calculating the future 

With the development of novel, innovative clinical trial design strategies, future biostatistics innovation will likely streamline drug discovery and clinical treatments. Tomi Mori, PhD, St. Jude Department of Biostatistics chair, sees the evolving role of biostatistics in clinical trials, emphasizing leveraging community efforts and collaboration to make the most of available data. She sees opportunities to utilize real-world data and innovative trial designs to accelerate drug development and improve patient outcomes.

"I believe there's a significant opportunity in biomedical informatics that we haven't fully tapped into, such as sharing molecular and clinical trial data," Mori explained. "If we can harness the power of all of the data to answer additional questions that cannot be answered in just one study, we can significantly contribute to the global effort in pediatric research."