The BCASA Third Annual Boston Pharmaceutical Symposium took place on May 3, 2019 at Pfizer’s building in Cambridge Massachusetts. Nearly 100 statisticians gathered to discuss statistical methodological research and its applications to biotech and pharmaceutical industry, and to develop connections. The participants had a wide variety of backgrounds, representing industry and academia, senior investigators, students and postdocs, and many were first-time attendees of BCASA events.

Weidong Zhang, the symposium’s host and a member of the symposium’s organizing committee kicked off the event by a short introduction, followed by two welcome addresses from Pfizer leadership team members. Then, 10 invited speakers representing Cytel, Biogen, Novartis, MD Anderson Cancer Center, Harvard University and Pfizer discussed topics on promising zone design, clinical development plans, drug-related safety signals, basket trial design, Bayesian optimal design, adaptive design in East 6.5, translational data science, co-data use in clinical trial, prognostic model construction by statistical learning method, meta-data for Emax model, etc. The audience had ample opportunities to interact during breaks, meet old colleagues and create new contacts.

BCASA thanks the symposium’s organizing committee (Olga Vitek, Northeastern University (chair); Weidong Zhang, Pfizer (host); Wenting Cheng, Biogen; Tim Clough, Novartis; Hrishikesh Kulkarni, Cytel; Andrew Lewandowski, Novartis; Jameson Luks, Cytel; Huyuan Yang, Alnylam Pharmaceuticals) for planning the event, as well as the Pfizer event team for hosting the symposium. We also extend our appreciation to the Bentley University student volunteers for the on-site coordinating work.

BCASA gratefully acknowledges financial support from Cytel.

Below is the event information.

Third Annual Boston Pharmaceutical Symposium

As an annual event, the Boston Pharmaceutical Symposium provides a unique venue for sharing statistical applications and research in the biotech-pharma industry, and building connections among all colleagues of the Greater Boston area engaged in the industry statistical practice. We welcome the participation from industry statisticians, academia researchers, as well as university students and any professionals who are interested in pharmaceutical statistical topics.

Third Boston Pharmaceutical Symposium was a full-day event, featuring a series of invited talks, a poster session, and networking opportunities.

Date and Time: Friday May 3, 2019, 8:00 am - 4:30 pm. Registration + light refreshments will start at 8:00 am. Welcome address will start at 8:45 am and talks begin at 9:00 am.

Symposium location: Pfizer, Building 2, Kendall Sq, 610 Main Street, Cambridge MA 02139 

Schedule of the Event

Morning session

Lunch and posters 12:00-1:00 pm

Afternoon session 

Scientific Committee

  • Olga Vitek, Northeastern University (chair) 
  • Weidong Zhang, Pfizer (host)
  • Wenting Cheng, Biogen
  • Tim Clough, Novartis
  • Hrishikesh Kulkarni, Cytel
  • Andrew Lewandowski, Novartis
  • Jameson Luks, Cytel
  • Huyuan Yang, Alnylam Pharmaceuticals


Abstracts and Bios

Craig Mallinckrodt, Distinguished Biostatistician, Biogen

Title: A practical guide to optimizing clinical development plans

Abstract: This presentation will focus on the seminal research and the resulting key principles that guide optimization of clinical development plans. We will consider which factors have the greatest influence on efficient drug development and how we can leverage this knowledge to optimize development.

Bio: Dr. Mallinckrodt is a Distinguished Biostatistician at Biogen where his responsibilities include various roles in early phase development along with portfolio analytics and management. Craig is a Fellow of the American Statistical Association and recently won the Royal Statistical Society’s award for excellence in the pharmaceutical industry. He has led a number of industry work groups and has published extensively on a variety of topics, including missing data, estimands, and various aspects of clinical development optimization.

Sam Hsiao, Senior Biostatistician, Cytel

Title: Optimal promising zone designs

Abstract: Clinical trials with adaptive sample size reassessment based on an unblinded analysis of interim results are a popular class of adaptive designs. Such trials are typically designed by prespecifying a zone for the interim test statistic, termed the promising zone, along with a decision rule for increasing the sample size within that zone. Mehta and Pocock (2011) provided some examples of promising zone designs and discussed several procedures for controlling their type-1 error. They did not, however, address how to choose the promising zone or the corresponding sample size reassessment rule, and proposed instead that the operating characteristics of alternative promising zone designs could be compared by simulation. Jennison and Turnbull (2015) developed an approach based on maximizing expected utility whereby one could evaluate alternative promising zone designs relative to a gold-standard optimal design. In this talk, we discuss results of Hsiao, Liu, and Mehta (2018) showing how, by eliciting a few preferences from the trial sponsor, one can construct promising zone designs that are both intuitive and achieve the Jennison and Turnbull (2015) gold-standard for optimality.

Bio: Sam is a Cytel senior biostatistician helping clients in multiple therapeutic areas to implement advanced statistical trial methodologies, simulate complex adaptive designs with specialized computing tools, perform interim monitoring, and develop analysis plans and other documentation for regulatory consideration. In academia, he's held positions as Director of the Mathematics Program at Bard College and NSF Postdoctoral Research Fellow at the University of Michigan. He's published in mathematical journals and collaborates on applied and methodological statistics with colleagues. Sam holds a PhD in mathematics from Cornell University and an MS in biostatistics from Harvard University.


Laurence Colin, Director, Novartis

Abstract: Nearly half of failed drug submissions can be attributed, at least in part, to safety issues. Better tools to identify drug-related safety signals early and with high accuracy would be useful. We will show how data from previous Novartis trials can be used to determine the likelihood that a safety signal observed in a first-in-human trial is related to the investigational compound under study.

Bio: Laurence Colin is a Director and group head in Translational Clinical Oncology at Novartis. In her 13 years at Novartis, she has primarily worked in early development, including as Disease Area Statistical Lead for autoimmune diseases and cardiovascular and metabolic diseases. Her areas of research interest include longitudinal modeling, dose-response analysis for safety endpoints, and analysis of large clinical databases. She has a MSc in Mathematics from University of Liege, Belgium, and a MSc in Biostatistics from the University of Hasselt, Belgium.


Matthias Kormaksson, Associate Director Statistical Consultant, Novartis

Title: A novel statistical learning method for building prognostic models for heart failure time-to-event outcomes and comparison to other machine learning approaches

Abstract: The original objective of this work was to assess performance of three survival analysis methods for predicting time-to-event in heart failure studies. The three methods were Cox regularized regression (Cox-LASSO), generalized additive Cox regression (Cox-GAM), and random survival forest, which provides a popular alternative from the machine learning community. The first two methods have the attractive properties of allowing for variable selection (Cox-LASSO) and modeling of non-linear features (Cox-GAM). However, currently there is no method that can incorporate both of these desirable properties. Therefore, we developed a novel hybrid method involving generalized additive Cox regression with additional L1-penalty (Cox-GAMLASSO). All models were trained and tested on data from two large studies in the Novartis Heart Failure database that includes both clinical and biomarker data. Model comparisons were made based on both discrimination and calibration performance. We conclude with lessons learned about the advantages and disadvantages of the different statistical and machine learning methods.

Bio: Matthias Kormaksson is a Principal Statistical Consultant in the Advanced Exploratory Analysis group at Novartis Pharmaceuticals, where he works on statistical machine learning problems in clinical trials, in particular on unsupervised clustering problems for identifying disease subtypes and supervised problems for response predictions. From 2012 to 2017, he worked as a research staff member and statistical consultant at IBM Research in Brazil, where his focus was on the application of spatial and spatio-temporal predictive models to problems arising in the area of natural resources. He received his Ph.D. in Statistics at Cornell University in 2010 and did a postdoc in the Division of Public Health and Epidemiology at Weill Cornell Medical School, where he did research involving microarray and next generation sequencing data arising in cancer epigenetics.

Chunlei Ke, Senior Director, Biogen

Title: Bayesian basket trial design

Abstract: A basket trial is a type of master protocol to study a single targeted therapy in the context of multiple diseases or disease subtypes (Woodcock and LaVange, 2017). If used properly, a basket trial can improve trial efficiency, and in some cases, may be the only feasible design option. Basket trial design has recently gained popularity particularly in oncology drug development due to great advancement in tumor biology and genomics. The characteristics of Bayesian design fit well into the features of Basket design and several Bayesian basket trial designs have been proposed in literature. We will first review some of these design proposals. Then we will introduce a new Bayesian adaptive basket design using a Bayesian model averaging framework. Its performance will be compared to that of some alternative design options. The proposed Bayesian model averaging framework can also provide a useful method to analyze data from a Basket trial, which will be illustrated with a real dataset.

Bio: Chunlei Ke is currently a Senior Director of Biostatistics at Biogen. He has worked in the field of clinical development for about 19 years across therapeutic areas of cardiovascular, oncology and neuroscience. His interest includes drug development strategy, novel trial design, and general statistical issues in clinical trials. He received his PhD degree in statistics in 2000 from University of California, Santa Barbara.

Ying Yuan, Professor, Department of Biostatistics, University of Texas, MD Anderson Cancer Center

Title: BOP2: Bayesian optimal design for phase II clinical trials with binary, co-primary and other complex endpoints

Abstract: The endpoints for immunotherapy and targeted therapy are often complicated, making conventional phase II trial designs or commonly used basket designs inefficient and disfunctional. We propose a flexible Bayesian optimal phase II (BOP2) design that is capable of handling simple (e.g., binary) and complicated (e.g., ordinal, nested and co-primary) endpoints under a unified framework. We use a Dirichlet-multinomial model to accommodate different types of endpoints. At each interim, the go/no-go decision is made by evaluating a set of posterior probabilities of the events of interest, which is optimized to maximize power or minimize the number of patients under the null hypothesis. Unlike most existing Bayesian designs, the BOP2 design explicitly controls the type I error rate, thereby bridging the gap between Bayesian designs and frequentist designs. In addition, the stopping boundary of the BOP2 design can be enumerated prior to the onset of the trial. These features make the BOP2 design accessible to a wide range of users and regulatory agencies, and particularly easy to implement in practice. Simulation studies show that the BOP2 design has favorable operating characteristics with higher power and lower risk of incorrectly terminating the trial than some existing Bayesian phase II designs. The software to implement the BOP2 design is freely available at www.trialdesign.org 

Neal Thomas, Executive Director, Pfizer

Title: Meta-data and software for Bayesian Emax dose response models

Abstract: Meta-analyses and an R package for clinical dose response that support a Bayesian model-based approach to designs and analyses will be described. Meta-data from approximately 200 compounds will be briefly summarized. With few exceptions, the meta-data support the Emax model commonly recommended by clinical pharmacologists. Meta-analyses also yield an empirical basis for specifying prior distributions for the parameters in the difficult-to estimate Emax model. We will conclude with an example utilizing the R package to demonstrate how the meta-analysis findings were applied to assess a proposed design for a recent study.

Hrishikesh Kulkarni, Principal Statistician, Cytel

Title: Overview of adaptive designs for confirmatory trials in East 6.5

Abstract: Adaptive clinical trials are becoming more common in clinical trials. This talk will be a high-level overview of several interesting additions to the latest version of East, v.6.5. It will include areas like using MCPMod for designing and analyzing Phase 2 studies, adaptive sample size re-estimation, prediction of events and enrollment at the planning as well as interim time point in a trial and lastly, more advanced methods like Population Enrichment.

Bio: Hrishikesh Kulkarni is a Principal Statistician and Customer Experience Manager at Cytel. He joined Cytel in 2007 and has been involved in the software quality, documentation, demos & training, and product marketing for East, EnForeSys and other Cytel products. Hrishikesh has M.Sc. in Statistics from University of Pune, India. He is based in Cambridge, MA.

Satrajit Roychoudhury, Senior Director, Pfizer

Title: On the use of co-data in clinical trial

Abstract: Historical data are important for the design of a clinical trial. Yet these data are rarely used in the analysis of the actual trial. While justifiable in certain situations, ignoring historical data can lead to less accurate inferences, and, therefore, suboptimal decisions. After a review of the main approaches to using historical data, the framework is extended to co-data, which comprise all relevant (historical and concurrent) trial-external data. These data can be used for the inference of the parameter in the actual trial via meta-analytic approaches. While the use of co-data in clinical trials is attractive, it is also ambitious. For example, avoiding undue weight of co-data (relative to actual trial data) is important, which can often be achieved by plausible assumptions about between-trial heterogeneity and allowance for nonexchangeability across trial parameters. Two applications with co-data will be discussed: phase III trials with interim decisions informed by co-data; and, a phase I combination trial in Oncology, which takes advantage of co-data from completed and ongoing phase I trials. 

Bio: Dr. Satrajit Roychoudhury is a Senior Director and a member of Statistical Research and Innovation group in Pfizer Inc. Prior to joining, he was a member of Statistical Methodology and Consulting group at Novartis. He started his career as a research statistician in Schering-Plough Research Institute (now Merck Co.). He has 10+ years of extensive experience in working in different phases of clinical trials. His primary expertise includes implementation of innovative statistical methodology in clinical trials. He has co-authored several publications/book chapters in this area and provided statistical training at major conferences. His areas of research include the use of survival analysis, model-informed drug development and Bayesian methods in clinical trials.

L.J. Wei, Professor of Biostatistics, Harvard University

Title: Translational data science

Abstract: A translational statistic is an empirical summary of data for making inferences about a population parameter of interest. Such statistic can be interpreted physically or clinically and comprehended heuristically for decision makings by, for example, clinicians and patients, on treatment selections. Like translational medicine, translational statistical research is to apply basic data science findings to the real-world clinical practice. Unfortunately, many commonly used statistics in clinical studies are not readily translatable. For example, a p-value does not have clinical interpretation. To make statistical research translational, we may estimate the parameter of interest empirically via robust and efficient statistical procedures. Ideally the parameter should be model-free and clinically interpretable. However, some commonly used estimation procedures may result in ambiguous, uninterpretable conclusions. For example, to evaluate an association between a biomarker and an outcome variable in the presence of various confounders, multivariate regression “working” models are routinely used to estimate the regression coefficient to quantify the strength of an association. The resulting estimates from different working models may estimate quite different parameters, which are likely not the original parameter quantifying the association. Therefore, it is difficult, if not impossible, to quantify the impact from this biomarker on the outcome variable. In practice, we still rely on the p-value to assess the association qualitatively. In fact, if the p-value for testing the regression coefficient being 0, is less than 0.05, one usually claims that this biomarker is an “independent predictor.” Such a claim is ambiguous and has very little value in clinical practice. In this talk, we will discuss several examples for illustration and show how to make conventional statistic research more translational.

Bio: L.J. Wei is a professor of Biostatistics at Harvard University. Before joining Harvard, he was a professor at University of Wisconsin, University of Michigan, and George Washington University. His main research interest is in the clinical trial methodology, especially in design, monitoring and analysis of studies. He has developed numerous novel statistical methods which are utilized in practice. He received the prestigious Wald Medal in 2009 from the American Statistical Association for his contribution to clinical trial methodology. He is a fellow of American Statistical Associating and Institute of Mathematical Statistics. In 2014, to honor his mentorship, Harvard School of Public Health established a Wei-family scholarship to support students studying biostatistics. His recent research area is concentrated on translational statistics, the personalize medicine under the risk-benefit paradigm via biomarkers and revitalizing clinical trial methodology. He has more than 180 publications and served on numerous editorial and scientific advisory boards. Dr. Wei has been closely working with pharmaceutical industry and the regulatory agencies for developing and evaluating new drugs/devices. 

Acknowledgements: We thank our colleagues at Pfizer for hosting this event. Financial support from Cytel is also gratefully acknowledged.