When determining the best possible statistical design for a particular trial, large pharmaceuticals and small biotechs benefit from different quantitative strategies. A new line of study from Cytel statisticians, with colleagues at Harvard University, Bristol Meyers Squibb, and Vertex Pharmaceuticals, reveals the tactical advantages of different approaches to sample size re-estimation. Using the familiar promising zone design, Cyrus Mehta, Pralay Senchaudhuri, Apurva Bhingare and Lingyun Liu demonstrate how to optimize for power, bearing in mind considerations like resource constraints and commercial goals.
What are Promising Zone Designs?
Promising zone designs refer to methods of adaptive sample size re-estimation, which enable sponsors to increase sample size based on whether the data gathered at an interim look is promising. The design enables statisticians to examine the interim data of an on-going trial and make an estimate as to whether the trial will be sufficiently powered if allowed to continue. A decision can then be made to either stop for futility, stop for overwhelming efficacy, or if deemed promising, can extend the sample size to secure more power.
How to Use Conditional and Unconditional Power Tactically:
There are at least two ways to predict the power of a clinical trial during the interim look.
Unconditional Power: Unconditional power is calculated by using the existing data to forecast the outcome for a large number of hypothetical future clinical trials similar to the present one , and then calculating the average success rate across those trials.
Conditional Power: Conditional power is calculated by using existing data to forecast whether or not the current trial will be sufficiently powered at the point of completion. Conditional power has a Bayesian analogue in predictive power.
Pharmaceutical companies with dozens of clinical trials in their portfolio are positioned to use unconditional power measures, since over the course of several trials the average success rate should equal the desired unconditional power. By contrast, biotechs pursuing a trial for a single asset, might value conditional power as a clearer guide on how to distribute clinical trial resources.
Which treatment effect assumptions should we use?
When designing clinical trials with sample size re-estimation, it is usually the case that sponsors assume a realistic (or often optimistic) treatment effect. Yet clearly there is value to knowing whether a sample size increase will facilitate detection of a smaller treatment effect that is both a clinically meaningful and economically viable. Mehta, et al., demonstrate that increasing the sample size to detect a pessimistic treatment effect can add tactical value to a design even though it comes at a cost to sacrificing a small amount of unconditional power. They propose two different sample size re-estimation rules (frequentist and Bayesian) that guarantee sufficient power to detect clinically meaningful treatment effects if the interim results are promising.
We develop optimal decision rules for sample size re-estimation in two-stage adaptive group sequential clinical trials. It is usual for the initial sample size specification of such trials to be adequate to detect a realistic treatment effect δ a with good power, but not sufficient to detect the smallest clinically meaningful treatment effect δ min . Moreover it is difficult for the sponsors of such trials to make the up-front commitment needed to adequately power a study to detect δ min . It is easier to justify increasing the sample size if the interim data enter a so-called “promising zone” that ensures with high probability that the trial will succeed. We have considered promising zone designs that optimize unconditional power and promising zone designs that optimize conditional power and have discussed the tension that exists between these two objectives. Where there is reluctance to base the sample size re-estimation rule on the parameter δ min we propose a Bayesian option whereby a prior distribution is assigned to the unknown treatment effect δ, which is then integrated out of the objective function with respect to its posterior distribution at the interim analysis.
About the Author of Blog:
Dr. Esha Senchaudhuri is a research and communications specialist, committed to helping scholars and scientists translate their research findings to public and private sector executives. At Cytel Esha leads content strategy and content production across the company's five business units. She received a doctorate from the London School of Economics in philosophy, and is a former early-career policy fellow of the American Academy of Arts and Sciences. She has taught medical ethics at the Harvard School of Public Health (TH Chan School), and sits on the Steering Committee of the Society for Women in Philosophy's Eastern Division, which is responsible for awarding the Distinguished Woman in Philosophy Award.