The current wave of clinical trials for developing personalized—or precision—medicine, in cancer and other diseases, faces a major challenge: patient accrual can be long and difficult. This is because only participants with a specific biological feature, which may be a gene mutation, are qualified in a precision medicine trial. This has been recognized as a problem for years, especially for rare cancers in which finding enough patients with a specific gene signature can take decades. Dr. Gang Han, associate professor at the Texas A&M School of Public Health with a doctoral degree in statistics, has developed statistical methods to better model survival outcomes to understand the new treatment benefits and better design such trials.
[Photo: Dr. Gang Han]
Many clinical trials testing a new cancer therapy measure success by how many people in the study are alive at a certain point in time—two years, five years or 10 years later. “When answering yes or no to the question about survival at a certain point, we are treating the outcome as dichotomous although the measurement of time to disease progression or time to death is continuous. Dichotomizing a continuous outcome is similar to throwing one third of the patient information away,” Dr. Han said. “Our approach can take into account the distribution of the time to an event. We propose to use a flexible parametric survival estimate for the whole distribution using all data information.” Dr. Han and colleagues have been developing this reduced piecewise exponential analytical framework for more than eight years to facilitate precision cancer research.
This may enable clinicians to design a trial with fewer participants, routinely saving one-third of patients compared with existing commonly used statistical methods. To both clinicians and patients, the benefit of Dr. Han’s work could be critical. “If a novel treatment has significant effects, we can quickly identify the effects and design new trials for next phase,” Dr. Han said. “If it is not showing significant effects, we can quickly reject it to save time and cost. Either way, using the novel analytical method is ethically attractive and cost effective.”
More specifically, in a single-arm trial where all participants receive a new treatment, Dr. Han’s method can better estimate the survival probabilities at landmark times as well as other survival measurements: for example, median and mean survival time. In a two-arm or multi-arm trial where two or more therapies are being compared, Dr. Han’s method can better capture the signal of improvements from the novel treatment. “Compared with existing treatments, additional benefits from a new treatment may not be noticeable in the first few weeks or even few months. But in another trial, the additional benefit may only be significant in the first few weeks or months,” Dr. Han said. “As a result, comparing treatments for the whole trial period will miss the signal. Our method involves a statistical test identifying time points where the survival risk changed significantly for each treatment arm. The comparison of the corresponding time periods can show additional benefits from the new treatment, leading to more efficient and effective risk stratification and disease monitoring in practice.”
Trained as a mathematical statistician, Dr. Han has published many peer reviewed articles on the reduced piecewise exponential framework and has implemented the statistical methods he developed in a number of clinical and pathological studies. Dr. Han has received more than 20 research grants with health science researchers, and some of the research grants were used to block his time for methodological research. Dr. Han will spend the next year working on funded projects on lung cancer, melanoma, and aging research, where the patient outcomes include overall survival, progression-free survival and recurrence-free survival. “I will implement appropriate statistical methods in practical health science studies,” Dr. Han said, “and I will continue to develop new methods to solve analytical problems that arise along with the current trend of precision medicine.”Tags: Texas A&M