AM Short course: (9:00AM – 12:00PM local time, 6/27)
Title: Statistical and Strategic Perspectives for Decision-making in Early and Early-to-Late Transition Oncology Drug Development
Instructors: Cong Chen (Merck); Cindy Lu (AstraZeneca)
Abstract: Building on the unprecedent success of immune checkpoint inhibitors in oncology drug development during the past decade, an expanding wave of promising mechanisms of action and innovative modalities has surfaced, each holding the potential for groundbreaking advancements. Amidst these scientific innovations, it is more imperative than ever for the drug developers to make swift and informed investment decisions. A framework of decision making based upon quantitative and strategic assessments of the cumulative evidence is essential to guide our forthcoming decisions effectively.
In this short course, we will talk about decision making post dose-finding from both statistical and strategic perspectives, as well as development considerations on combination therapies of an investigational new drug with standard-of-care. Along with this journey, we will partition the short course into three parts:
Part I: General decision making framework post dose-finding. In this part, we will discuss the general decision making framework under the challenges of small sample sizes. Decision making at reflection points such as from dose escalation to dose expansion, and post dose expansion before phase III investment decisions are considered. Handling different endpoints at those reflection points will be discussed with real demonstrative examples.
Part II: Decision making from benefit-cost ratio (BCR) analysis perspective. It shows that, while prior data on drug activity is crucial for determining the aggressiveness of a Phase 2/3 program (e.g., Phase 2, adaptive Phase 2/3, or Phase 3), once the program is launched, it becomes less important than the relative cost savings in deciding whether to continue. It is often unknown whether a drug has the target treatment effect, but projecting costs is easier, allowing for the creation of robust baseline decision rules based on BCR analysis. While Type I and II errors are often a focus at the trial level, Type III error is a concern at the portfolio level. Properly balancing all three is key to long-term success, which can be similarly achieved with BCR analysis.
Part III: 2-in-1 Design and independent drug action in combination therapies. In this part, we will introduce the 2-in-1 design and its statistical properties and strategic implications. We will also discuss independent drug action model, based on which an innovative approach to predict the overall response rate, progression-free survival and mean response duration of combination therapies will be introduced, followed by discussions on its statistical implications for trial design and monitoring. Further considerations on contributions of components in combination therapies will also be cover in this part.
Expected outcomes: Attendees of this short course will not only learn technical skills that can be immediately applied in practice, but also gain an enhanced perspective of strategic decision-making in oncology drug-development.
Target audience are biostatisticians working in the biopharmaceutical industry, especially in oncology drug development.
Brief Bios for the Instructors:
Dr. Cong Chen is Scientific AVP in Early Development Statistics at Merck & Co., Inc., providing fit-for-purpose decision-making strategies and novel statistical approaches for early and early-to-late transitional oncology programs, and supporting oncology external collaborations, competitive intelligence, and high-profile due diligence projects. Prior to taking the role, he led the statistical support for the development of pembrolizumab and played a key role in accelerating its regulatory approvals. He is an elected Fellow of American Statistical Association (2016), an Associate Editor of Statistics in Biopharmaceutical Research, a member of Cancer Clinical Research Editorial Board and a leader of the DahShu Innovative Design Working Group. He has published over 100 papers and 10 book chapters on innovative design and analysis methods of clinical trials, has given multiple short courses on the subject at statistical conferences and was invited to give oral presentations at multiple medical conferences on design strategies for oncology drug development.
Cindy Lu, Ph.D. is a senior director of Oncology Biometrics in AstraZeneca. She focused her over 15 years of career in drug discovery and development, led statistical aspects for multiple compounds across various disease areas, from early to late phases of clinical development to post-marking activities, resulting in several successful regulatory approvals. Dr. Lu is currently leading multiple scientific working groups within American Statistical Association (ASA) and DahShu on innovative design and oncology statistical methodology. In her current role in AstraZeneca, she is supervising a group of statisticians overseeing a broad variety of early phase oncology assets, where constant qualitative decision-makings are required to expediate drug development in bringing innovative therapies to patients.
PM Short course: (2:30PM – 5:30PM local time, 6/27)
Title: An Introduction to Sufficient Dimension Reduction
Instructor: Bing Li
Abstract: In this short course I will trace the development of Sufficient Dimension Reduction for the past three decades or so. From linear dimension reduction to nonlinear dimension reduction, and from Euclidean data, to functional data, to tensorial data, to random objects. I will also review the functional classes that are used to develop the wide class of sufficient dimension reduction estimators: linear functions in Euclidean spaces, nonlinear functions in reproducing kernel Hilbert spaces, and nonlinear functions in neural network classes.
Brief Bio for the Instructor:
Bing Li is Verne M. Willaman Professor of Statistics at the Pennsylvania State University. He is a fellow of ASA and IMS. He served as an associate editor for the Annals of Statistics and for JASA. He authored the book "Sufficient Dimension Reduction: Methods and Applications with R" and coauthored the textbook "A Graduate Course on Statistical Inference". His research interests include sufficient dimension reduction, statistical graphical models, functional data analysis, metric-space-valued data analysis, kernel learning, deep learning, asymptotic analysis, estimating equations, longitudinal data, and robustness.