Online Program

Search by Last Name:



Mon, Sep 20

SC1 Emerging Challenges in Clinical Trial Methodologies

8:30 AM - 12:00 PM
Constitution Room B

Instructor(s): Hsien (James) Hung, FDA; Sue-Jane Wang, FDA

In the last decade, the clinical trial methodology involved in regulatory applications is increasingly complex. As the awareness of potential ethical issues with use of placebo increases, an active control is increasingly considered as a comparative arm for assessing the effect of a test treatment. Non-inferiority trial design has therefore been revisited. On another front, even for the placebo-controlled trial, the conventional standard design has been faced with many challenges aiming at design adaptability, following the invention of group sequential designs. Evidential standard promotes consideration of more innovative trial designs and inferential frameworks for studying more than one objective or endpoint in a regulatory submission. This short course will discuss some emerging challenges in the following topics: active controlled trial design, adaptive design, multiple comparison considerations, roles of modeling and simulation, pharmacogenomics.


SC2 Beyond Survival Analysis: Recurrent Event Responses in Clinical Trials

8:30 AM - 12:00 PM
Constitution Room A

Instructor(s): Richard Cook, University of Waterloo; Jerry Lawless, University of Waterloo

The aim of this short course is to introduce models and statistical methods available for the analysis of recurrent event data in clinical trials. The emphasis will be on models with multiplicative rate functions and robust variance estimation. Features of the various approaches will be illustrated by application to data from recent clinical trials and techniques for model assessment will be discussed. Issues arising due to mortality will be discussed along with considerations at the planning stages of trials. Statistical analysis will be carried out using SAS and R/S-PLUS code. The course will be based on material from the book The Statistical Analysis of Recurrent Events (Springer, 2007) written by the presenters.


SC3 Interpreting Change and Responder Analysis for Patient-Reported Outcomes

8:30 AM - 12:00 PM
Constitution Room C

Instructor(s): Joe Cappelleri, Pfizer; Lisa A. Kammerman, FDA; Kathleen W. Wyrwich, UnitedBioSource Corporation

Patient-reported outcome (PRO) measures used for labeling and promotional claims must have: 1) evidence documenting their responsiveness; and 2) interpretation guidelines (e.g., responder definition) to be most useful as effectiveness endpoints in clinical trials. The recommended approach is to estimate the responder definition based on anchor-based methods, which will be discussed during the workshop. However, this workshop will also discuss how distribution-based methods can provide some insights on interpreting the amount of change that signifies an important change in PROs. Confidence in a specific responder change threshold evolves over time and is confirmed by additional research evidence, including clinical trial experience; the responder change threshold may vary by population and context, and no one responder change threshold will be valid for all study applications involving a PRO instrument. During this workshop, the speakers will explain how to demonstrate and identify thresholds for specific study populations in an effort to pursue labeling and promotional claims.


SC4 Hot Topics: Recent Development in Clinical Trial Methodologies

1:30 PM - 5:00 PM
Constitution Room B

Instructor(s): Scott Evans, Harvard University; Jim Ware, Harvard University

This course addresses four hot topics in clinical trials: noninferiority (NI) trials, subgroup analyses, benefit:risk (B:R) assessment, and interim data monitoring. The presentations include motivating examples, discussion of challenging issues, and novel approaches to these issues. The NI module includes a discussion of the challenges of design, analysis, interpretation, and methods for defining the NI margin and testing for NI. The subgroup analysis module includes a discussion of the rationale for the guidelines for the reporting of subgroup analyses adopted by the New England Journal of Medicine. The B:R module highlights novel approaches to personalized medicine (tailored treatment selection). The interim data monitoring module will include novel approaches to the analysis and presentation of interim data to DMCs including the use of prediction in adaptive design.


SC5 Good Statistical Practice and Common Subtle Statistical Mistakes

1:30 PM - 5:00 PM
Constitution Room A

Instructor(s): Frank E Harrell, Jr., Vanderbilt University

This workshop deals with principles derived from over 30 years of applying statistics to biomedical research, collaborating with clinical and basic biological researchers and epidemiologists. The principles relate to statistical efficiency, bias, validity, robustness, interpretation of statistical results, multivariable predictive modeling, statistical computing, and graphical presentation of information. Topics to be discussed include respecting continuous variables, avoiding non-descriptive statistics, problems associated with filtering out negative results, overfitting, shrinkage, adjusting P-values for multiple comparisons without adjusting point estimates for same, and the false promise of multi-stage estimation and testing procedures, related to the use of bogus conditional techniques for computing what is advertised as unconditional variances or type errors.


SC6 Quantitative Pharmacovigilance: Statistical Approaches to Medical Product Safety Surveillance

1:30 PM - 5:00 PM
Constitution Room C

Instructor(s): Jie Chen, Merck Serono; Yi Tsong, FDA

With the issuance of the Sentinel Initiative, the FDA is aggressively seeking ways towards the modernization of the science of safety, which encompasses the entire life cycle of a product, from pre-clinical animal studies, early clinical human safety tests, late stage clinical development, to post-marketing widespread use in general population. The science of safety employs effective methods for surveillance, signal detection, data mining, modeling and analysis that are enabling researchers to generate hypotheses about and confirm the existence and cause of safety problem. Appropriate applications of statistical approaches and their integration into routine Pharmacovigilance practice will ensure early detection of true safety signals, reduce the rate of false signals and false non-signals, and hence protect the health of consumers as well as the manufacturers. This short course covers materials of statistical methods for identification, quantification, and evaluation of medical product safety signals in both clinical and post-marketing settings.


Tue, Sep 21

General Session 1 Regulatory Policy and Planning in Benefit-Risk Evaluation of Medical Products

8:45 AM - 10:15 AM
Constitution Ballroom

Organizer(s): Bruce Binkowitz, Merck and Co. Inc.; Scott Evans, Harvard University; Richard Kotz, FDA

Chair(s): Bruce Binkowitz, Merck and Co. Inc.

The basic and critical mission of FDA and industry is to ensure safe medical products which requires essential, effective, evidence-based assessments of risks and benefits. This session will take a systematic look at the philosophy, planning and policies in benefit-risk evaluation. Topics covered will include principles to consider, the need to balance safety-risk with benefits, identifying well defined criteria, and consideration of both qualitative and quantitative assessments, as well as encourage clear and effective communication. This session is part of a two-session series on benefit-risk evaluation on medical drug and device development. Speakers from regulatory, academia and industry will discuss challenges in benefit-risk evaluation and decision process as well as future directions from a regulatory and policy planning point of view.

Considerations in Evaluating Risks and Benefits of Medical Interventions
John H. Powers, NIH and School of Medicine, George Washington University and University of Maryland

Quantitative Benefit/Risk Assessment: Here and Now
View Presentation View Presentation Christy Chuang-Stein, Pfizer

Benefit Risk from a Regulatory Perspective
Joyce Korvick, FDA


General Session 2 Collective Evidence from Multiple Studies – Is Multiplicity Adjustment Needed?

10:30 AM - 12:00 PM
Constitution Ballroom

Organizer(s): Xiaoping Jiang, FDA; Qian Li, FDA; Tonya Marmon, Synteract; Brian Wiens, Alcon; Casey Xu, FDA

Chair(s): Qian Li, FDA

In regulatory decisions, the totality of evidence from multiple studies is evaluated collectively to provide sound evidence upon which to base the decisions. The collective evidence evaluates multiple studies, multiple arms, and multiple endpoints. The logic of the collective evidence is that if multiple endpoints and multiple arms all show evidence of efficacy consistently across multiple studies, the probability that the demonstrated effect is due to chance will be extremely small. This logic is the basis of the error probability of wrongly approving an ineffective medical product using collective evidence. On the other hand, controlling Type I error in each individual study isolates evidence and uses stringent multiplicity adjustment procedures. This approach often does not result in a sensible conclusion to support a regulatory decision and sometimes becomes an obstacle in decision making. Cases from FDA reviews are discussed in this session. Questions from statistical reviewers at the FDA are debated among a group of prestigious panelists from FDA, academia, and industry.

Applications of Collective Evidence in Reviews of Medical Products
Greg Soon, FDA

Replication, Consistency of Effect, and Control of the Overall Type 1 Error Rate
View Presentation View Presentation Steven Snapinn, Amgen/GBE

Bob Temple, FDA; Gregory Campbell, U.S. Food and Drug Administration; Greg Soon, FDA; Jon Norton, FDA; Mohammad Huque, FDA; Thomas Fleming, University of Washington; Steven Snapinn, Amgen/GBE


RT1 Roundtable Topic: Design and Implementation Trials

12:00 PM - 1:15 PM
Independence BCDE

#1a: Site and Central Imaging Evaluations: Bias and Variance in Study Design and Analysis
David Raunig, Pfizer; Kohkan Shamsi, RadMD

#1b: Evaluating Multinational Clinical Trials
Shiling Ruan, FDA

#1c: Alzheimer’s Disease Designs
Xiaoling Wu, BMS

#1d: Bayesian Clinical Trials
Xiting (Cindy) Yang, FDA

#1e: Study Design and Analysis of Genomic Data in Clinical Studies
Zhaoling Meng, Sanofi Aventis

#1f: Current Challenges and Key Statistical Elements in the Design of Phase 2 Dose-Response Studies
Qing Liu, Johnson & Johnson; David Petullo, FDA

#1g: Logistics and Implementation of Adaptive Trials
Eva Miller, ICON

#1h: Composite Endpoints
Hsien (James) Hung, FDA;

#1i: Statistics of Early Dose-Determination: Ad-Hoc Versus Truly Adaptive
William Mietlowski, Novartis; Sue-Jane Wang, FDA

#1j: Determining Guidelines to Assist DMC/DSMB Stopping Decisions for Safety Concerns at Interim Analyses
Paul Gallo, Novartis Pharmaceuticals; Sandeep Menon, Biogen Idec

#1k: Considerations in Obesity Clinical Trial Designs
Shailaja Suryawanshi, Merck & Co., Inc.

#1l: Sample Size Estimation for Non-inferiority Clinical Trials
Shaoyi Li, Celgene Corporation

#1m: Statistical and Regulatory Challenges of International Trials
Bruce Binkowitz, Merck and Co. Inc.; Richard Kotz, FDA


RT2 Roundtable Topic: Analysis of Clinical Trials (Safety and Efficacy)

12:00 PM - 1:15 PM
Independence BCDE

#2a: Site by Treatment Interactions
Christine Blasey, Corcept Therapeutics Inc.

#2b: Statistical Issues in Meta-Analysis of Drug Eluting Stent Data
Hsini (Terry) Liao, Boston Scientific

#2c: The Benefit of Stratification in Clinical Trials Revisited
Jitendra Ganju, Amgen

#2d: Challenges in the Analysis of Correlates of Protection in Vaccine Trials
Mark Wolff, Emmes; Lihan Yan, FDA

#2e: Problems Associated with Unequal Center Size
Brian J. Fergen, USDA, CVB; Louis Luempert III, Novartis Animal Health

#2f: Estimating Active Control Effects and the Choice of Non-Inferiority Margin
Mark Rothmann, FDA; Brian Wiens, Alcon

#2g: Statistical Graphics for Clinical Data Analysis
Michael O'Connell, Spotfire, TIBCO Software, Inc

#2h: Analysis of Poolability of Sites
Pablo Bonangelino, FDA

#2i: Analysis of Extension Studies at Interim for a Submission and at Study Close-Out
John Jones, UCB

#2j: Exact Methods for safety analyses of clinical trials
Paul Schuette, FDA

#2k: Presentation of Multivariate Modeling for Clinical Use
Shannon Song, Boston Scientific


RT3 Roundtable Topic: Center-Specific Topics

12:00 PM - 1:15 PM
Independence F

#3a: Design and Analysis Issues in studies for the Validation and Verification of Performance Parameters of Diagnostic Assays
Marina Kondratovich, FDA; Robert Magari, Beckman Coulter Inc.; Jeff Vasks, Roche

#3b: Statistical Safety Data Analysis in Medical Device Clinical Trials and Post-Marketing Surveillance
Chang S. Lao, FDA

#3c: Diagnostic Device Submissions and the pre-IDE Process at FDA
Patrick Meyers, Abbott Laboratories; Estelle Russek-Cohen, FDA

#3d: Using Quantitative Decision Analysis in the Regulation of Medical Devices
Jason Schroeder, FDA


RT4 Roundtable Topic: CMC/Early Phase/Pre-Clinical

12:00 PM - 1:15 PM

#4a: Statistical Issues in the Analysis of Thorough QT/QTc Studies
Yi Tsong, FDA

#4b: Minimal Statistical Standards for Biomarker Assay Development (MiSSBAD)
Maha Karnoub, Wyeth Pharmaceuticals; Gene Anthony Pennello, Food and Drug Administration

#4c: How to Evaluate Diagnostic Biomarkers in a Regulatory Setting
Kyunghee Song, FDA


RT5 Roundtable Topic: Other Statistical/Technical Topics

12:00 PM - 1:15 PM
Independence BCDE

#5a: Tools and Software for Performing Some Complex Power Calculations Beyond the Two-Group Single Endpoint Design
Bob Abugov, FDA; Tad Archambault, Virtu Stat, Ltd.

#5b: CDISC: Statistician's Perspective
Tonya Marmon, Synteract; Behrang Vali, FDA

#5c: Off Shoring Experiences: What Works and What Doesn’t
Leonard Oppenheimer, Johnson & Johnson; Weiying Yuan, Johnson & Johnson

#5d: Biosimilarity: The Current Status from Regulatory, Clinical and Statistical Point of View
Lily Zhao, Novartis Oncology

#5e: Standardization and Qualification of (Genomic) Biomarkers
Nandini Raghavan, Johnson & Johnson Pharmaceutical Research & Development; Atiar Mohammed Rahman, FDA

#5f: Signal Detection for FDA's Safety Data
Lele Chitra, Sciformix; Ana Szarfman, FDA

#5g: Uses and Advantages of Open Source Software in the Pharmaceutical Industry and for FDA Statistical Reviewers
Frank E Harrell, Jr., Vanderbilt University; Bill Pikounis, Johnson & Johnson

#5h: Open Source Software in a Regulatory Setting
Keaven Anderson, Merck & Co., Inc.

#5i: Experience with SDTM/ADaM Dataset Submission Among Statisticians
Kalyan Ghosh, BMS; Joy Mele, FDA; Greg Soon, FDA

#5j: The Role of Simulations in FDA Submissions
Mark Chang, AMAG Pharmaceuticals; Jack Jie Zhou, FDA

#5k: Experience in Selecting, Installing and Using an Electronic Data Capture System (EDC) in Support of FDA Regulated Trials
Sunni A. Barnes, Baylor Health Care; Elisa Priest, Baylor Health Care


RT6 Roundtable Topic: Professional/Personal Development

12:00 PM - 1:15 PM
Independence G

#6a: Personalized Medicine, industry survey, company interviews
Christopher-Paul Milne, Tufts University

#6b: Symbiosis Between Sponsor and Contract Research Organization Statisticians
Nfii K. Ndikintum, Paragon Biomedical Inc.; Eshetu T. Wondmagegnehu, Eli Lilly and Company

#6c: FDAAA Webresults Disclosure
JoAnn Shapiro, Bayer

#6d: Sponsor/Independent Statistical Center: Establishing A Trustworthy Interaction for DMCs
William Coar, Axio Research


PS1a Issues and Methods for Handling Missing Data in Clinical Trials

1:15 PM - 2:30 PM
Constitution Room A

Organizer(s): G. Frank Liu, Merck & Co., Inc.; Satish Misra, FDA; Antonio Paredes, FDA; JoAnn Shapiro, Bayer; Greg Soon, FDA

Chair(s): Greg Soon, FDA

Missing data often impacts trial quality and excessive missing data may lead to inconsistent sensitivity analyses which could call into questions the validity of any conclusions to be drawn from the study. Therefore, it is important to carefully consider the study design, data collection, and patient retention plan to minimize missing data problem or its impact in supporting meaningful and valid conclusions. It is also essential to choose the proper endpoints and select appropriate statistical analysis methods as well as cover a reasonable range of sensitivity analyses. Further, the considerations of missing data could differ by design; for example, in non-inferiority vs. superiority trials. In this session, speakers from the US FDA, pharmaceutical industry, and academia will present the current thinking on missing data issues and up-to-date statistical methods for analyses of clinical trials with missing data, including discussions on trial design considerations, assumptions and sensitivity analyses.

Missing Data in Clinical Trials: Report of an NAS Panel
View Presentation View Presentation Roderick J Little, University of Michigan

Analysis of Clinical Trials with Missing Data: Robust Alternatives to Standard Methods
View Presentation View Presentation Devan Mehrotra, Merck Research Laboratories

Missing Data: NAS and FDA
View Presentation View Presentation Tom Permutt, FDA

Discussant(s): Hsien (James) Hung, FDA


PS1b Issues and Challenges Related to Responder Analyses

1:15 PM - 2:30 PM
Constitution Room B

Organizer(s): Yongman Kim, FDA/CDER; Nelson Lu, FDA/CDRH; Rajesh Nair, FDA/CDRH; Steven Snapinn, Amgen/GBE

Chair(s): Rajesh Nair, FDA/CDRH

The use of responder analysis has been growing in popularity on account of the need to establish clinical as well as statistical significance in clinical trials, and the perception that responder analysis accomplishes this. When the measurement of interest is on a continuous scale, clinical trials usually report group responses such as the mean treatment effect; in contrast an analysis in terms of responder rates is seen by some to be easier to interpret and to more directly capture the benefit to the individual. However this comes at a cost in statistical efficiency. This session will address some of the statistical issues involved in the analysis of responder data in clinical trials.

Responder Analysis 1 (Bad) and 2 (Good)
Janet Turk Wittes, Statistics Collaborative, Inc.

Issues and Challenges in Dichotomizing Continuous Variables in Clinical Trials
View Presentation View Presentation Qi Jiang, Amgen Inc.

Responder Analysis to Assist Decision Making: A Case Study in Chronic Pain Trials
David Petullo, FDA


PS1c Use of Clinical Trial Simulations to Maximize Success of Drug Development Programs

1:15 PM - 2:30 PM
Constitution Room C

Organizer(s): Zoran Antonijevic, Quintiles, Inc.; Elizabeth Kumm, Health Genome Sciences, Inc.; Caiyan Li, FDA; Chenguang Wang, FDA

Chair(s): Zoran Antonijevic, Quintiles, Inc.

There is an increased pressure on Pharmaceutical companies to improve the productivity, quality, and cost-effectiveness of their product development, and consequentially there is less reliance on “gut feeling” at decision points, with more focus on quantitative decision making. Due to complexity of drug development clinical trial simulations are becoming an essential tool to evaluate product development scenarios, and to facilitate decision making. The first talk by Dr. Chang will give an overview of application of simulations to various aspects and stages of drug development. This talk will also describe a number of different simulation techniques. The second talk by Dr. Pinheiro will focus on use of simulations for evaluating a range of product development strategies that will be compared based on the associated probability of success and expected net present value. In her discussion Dr. Wang will address simulations’ methods and applications from adaptive design to modern protocol design for drug development.

Monte Carlo Simulation for the Pharmaceutical Industry
View Presentation View Presentation Mark Chang, AMAG Pharmaceuticals

Using Probability of Program Success and Expected Net Present Value to Compare Alternative Development Strategies: A Case Study from the PhRMA Adaptive Dose-Ranging Studies WG
View Presentation View Presentation Jose Carlos Pinheiro, Johnson & Johnson PRD

Use Simulation Tools to Compare Empirical Models in Confirmatory Trials
View Presentation View Presentation Fanhui Kong, FDA


PS1d Challenges and Solutions in Design and Statistical Analysis for Diagnostic Medicines for Diagnostic Medicines

1:15 PM - 2:30 PM
Constitution Room D

Organizer(s): Meijuan Li, FDA; Patrick Meyers, Abbott Laboratories; Alicia Toledano, Statistics Collaborative, Inc.; Chava Zibman, FDA

Chair(s): Meijuan Li, FDA

This session will present current topics in the evaluation of medical diagnostic equipment. Its primary focus will be on methods to evaluate efficacy of specimen assay and imaging tools which continue to be of the highest importance in the effort to improve the quality of health care. Speaker one will discuss methods to handle equivocal (grey zone or undetermined) results in assay agreement and clinical specificity and sensitivity evaluations. Speaker two will discuss the challenges in high-dimensional classifiers used in diagnostics, some solutions will be outlined. Speaker three will discuss issues and suggestions for design and statistical analysis of a precision an study for continuous and semi-quantitative immunohistochemistry (IHC) assays. Speaker four will discuss the analysis that generalizes binary truth to multi-level truth (agreement, concordance, prediction probability) using computer-aided devices

Gray Zone Matters
View Presentation View Presentation Kristen Meier, FDA, CDRH; Vicki Petrides, Abbott Diagnostics

Diagnostics in High Dimension
View Presentation View Presentation Douglas Hawkins, University of Minnesota

Statistical Points to Consider in the Precision Testing of Immunohistochemistry Assays
View Presentation View Presentation Gene Anthony Pennello, Food and Drug Administration

Agreement by concordance and reader studies
View Presentation View Presentation Brandon D Gallas, FDA/CDRH/OSEL/DIAM


PS1e Risks in Your Everyday Food: Signal Detection for Global Food Safety

1:15 PM - 2:30 PM
Constitution Room E

Organizer(s): Yoko Adachi, U.S. Food and Drug Administration Center for Veterinary Medicine; Errol Strain, U.S. Food and Drug Administration Center for Food and Applied Nutrition

Chair(s): Yoko Adachi, U.S. Food and Drug Administration Center for Veterinary Medicine

Ensuring the safety of food in the era of globalization is no easy task, as illustrated by the example of melamine contamination in pet food and infant formula, and in the difficulty of importing/exporting food due to differing regulations on food additives, biotech food, and drug and chemical residues in meat and seafood. Furthermore, signal detection for food-related adverse events is substantially different from detecting adverse events in medical products as there is nothing analogous to prescription records and the FDA receives notification for only a small fraction of the events. Dr. Serratosa will discuss "Risk assessment approaches on Food safety in EU and US". Dr. Olafsson will discuss "Data Mining for Recognizing Patterns in Foodborne Disease Outbreaks". Dr. Chirtel will discuss recent developments at CFSAN. Dr. Bartholomew will give an overview of the structure of FDA and how we go about ensuring food safety. This session will address statisticians’ contributions to the surveillance effort through data mining and signal detection.

Risk Assessment in Food safety in the European Union, links with US
View Presentation View Presentation Jordi Serratosa, European Food Safety Authority

Data Mining for Recognizing Patterns in Foodborne Disease Outbreaks
Sigurdur Olafsson, Iowa State University Department of Industrial and Manufacturing Systems Engineering

Data Mining and Signal Detection in CFSAN's Adverse Event Reporting System
View Presentation View Presentation Stuart Jay Chirtel, FDA\CFSAN\OFDCER

Discussant(s): Mary Bartholomew, U.S. Food and Drug Administration Center for Veterinary Medicine


PS2a Statistical Methods in Benefit-Risk Evaluation of Medical Products

2:45 PM - 4:00 PM
Constitution Room A

Organizer(s): Carmen Mak, Merck & Co., Inc.; Jon Norton, FDA; Shiling Ruan, FDA

Chair(s): Carmen Mak, Merck & Co., Inc.

It is widely agreed that regulatory decisions should be based on an assessment of the potential benefits and harms from a product. There is no universally-accepted framework for this, nor is there a consensus about which method is suitable in which situation. This session is part of a two-session series on benefit-risk assessment. The purpose of this session is to get beyond the philosophical disputes and share quantitative methods for benefit-risk assessment. Presentations from industry, academic and regulatory experts will focus on practical advice with case study applications. Dr. Evans (Harvard) will speak on methods for personalized medicine and within-patient analyses of benefits and risks. He will also discuss the design, conduct, and reporting of clinical trials, as well as ideas for the future. Dr. Irony (FDA) will discuss the use of quantitative decision analysis in the regulation of medical devices. Dr. Cross (Genentech) will discuss health outcomes modeling as a way of quantifying the benefit-risk tradeoff of an intervention, going beyond traditional pre-approval clinical programs. He will also apply the methodology in an assessment of the benefit-risk profile of rosiglitazone (AvandiaTM) vs. comparators.

Personalized Medicine in Benefit:Risk Assessment
Scott Evans, Harvard University

Using Decision Analysis to Regulate Medical Devices
View Presentation View Presentation Telba Irony, CDRH, FDA

Health Outcomes Modeling to Quantify Benefit-Risk Tradeoffs: A Case Study Using Rosiglitazone
View Presentation View Presentation James Cross, Genentech/School of Pharmacy, University of Washington

Discussant(s): Ram Suresh, Merck & Co., Inc.


PS2b Multi-Regional Clinical Trials, Consistency, Sample Size

2:45 PM - 4:00 PM
Constitution Room B

Organizer(s): Daphne TY Lin, FDA; Kooros Mahjoob, FDA; Shailendra Menjoge, Boehringer-Ingelheim.; S. Peter Ouyang, Celgene

Chair(s): Shailendra Menjoge, Boehringer-Ingelheim.

Many statistical issues related to the conduct of multi-regional clinical trial for drug development were identified during the 2007 FDA/PhRMA meeting on the ”Challenges and Opportunities of Multiregional Clinical Trials”. A systematic review of existing approaches assessing consistency of treatment effects in a multi-regional clinical trial was presented in a plenary session of the 2009 FDA/Industry Statistics Workshop. Advancements are also being made on the following topics: definition of region, handling of different endpoints required by different regulatory authorities, statistical interpretation of the overall conclusion with or without a satisfactory regional consistency, explanatory analysis to account for apparent regional difference in treatment effect, and innovative design and analysis to allow additional bridging data when regional consistency is insufficient to draw region-specific conclusion. This session will feature new developments on these topics. Their applicability and ramifications will be discussed. Examples of multi-regional differences in the schizophrenia trials will be presented.

'Trial Design Issues and Treatment Effect Modeling in Multi-Regional Schizophrenia Trials
View Presentation View Presentation Yeh-Fong Chen, FDA

Recent Development of Statistical Methodology for Multi-Regional Clinical Trials
View Presentation View Presentation Xiaolong Luo, Celgene Corporation

Issues in implementing statistical evaluations of regional consistency
View Presentation View Presentation Paul Gallo, Novartis Pharmaceuticals


PS2c Vaccine Safety, H1N1 and Other Influenza Vaccines

2:45 PM - 4:00 PM
Constitution Room C

Organizer(s): Allen Izu, Novartis; Jingyee Kou, FDA/CBER

Chair(s): Jingyee Kou, FDA/CBER

Analysis of adverse events (AE) in phase 3 clinical trials is complicated by two conflicting facts: one of which is the low power for flagging any specific AE, and the other is the high rate of false positive findings. The first speaker will present a new analysis approach motivated by ideas contained in conditional power and false discovery rates with the goal of mitigating the influence of the conflicting factors. It will be followed by the industry story on swH1N1 development, and the FDA side on licensing influenza vaccines including swH1N1. Both speakers will cover the internal and external challenges (to the company or FDA) faced in the rapid development and licensing of the swH1N1 vaccine(s), and the lessons learned, particularly with respect to the demonstration of efficacy and safety

Adverse Event Signal Detection: Overall comparisons, Future Projections and False Discoveries
View Presentation View Presentation Jitendra Ganju, Amgen

The 2009 swH1N1 Pandemic - One Company's Experience and Lessons Learned
Klaus Stohr, Novartis Vaccines & Diagnostics

Influenza Vaccine Licensure: Are We Ready for the Next Pandemic or New Influenza Vaccines?
Tsai-Lien Lin, CBER, FDA


PS2d Biologics and Their Follow-on’s

2:45 PM - 4:00 PM
Constitution Room D

Organizer(s): Eric Chi, Amgen Inc.; Stella C. Grosser, FDA; Yi Tsong, FDA

Chair(s): Shein-Chung Chow, Duke University

A regulatory pathway for follow-on biologics (FoBs) needs to be established well before the first innovative biologic product gets out of patent. Companies that wish to register FoBs would then have guidance to follow and ultimately gain approval. The medical community will benefit from these FoBs as long as the regulatory pathway is scientific, patient safety focused, and at the same time, encourages innovation. Scientific background on the complexity of biologics, and how different they are from the small molecules will be discussed by Emily Shacter from the FDA. Jason Liao from Merck will discuss some scientific and statistical challenges in developing follow-on biologics and also present some thoughts on PK and PD comparability assessment. Finally Professor Laszlo Endrenyi from the University of Toronto will go over some work on assessing biosimilarity in variability of FoBs since the clinical outcomes of FoBs are very sensitive to slight variation in their structures and manufacturing processes.

Scientific considerations for the evaluation of Follow-on Biologics – FDA perspectives
View Presentation View Presentation Emily Shacter, FDA/CDER/OPS/OBP/DTP

Biosimilarity for Follow-on Biologics
View Presentation View Presentation Jason Liao, Merck Research Labroatories

An Evaluation of the Similarity of Follow-on Biologics
View Presentation View Presentation Laszlo Endrenyi, University of Toronto


PS2e Design and Analysis Issues in Active Control Veterinary Clinical Trials

2:45 PM - 4:00 PM
Constitution Room E

Organizer(s): Jing Han, FDA/CVM; Louis Luempert III, Novartis Animal Health; Anna Nevius, FDA/CVM

Chair(s): Jing Han, FDA/CVM

Veterinary trials present unique statistical challenges in the choice of control group and in adhering to the general principles for designing and conducting valid non-inferiority studies. This session will provide a forum for discussion of these issues from both industry and regulatory perspectives. Examples will be presented to demonstrate the opportunities and challenges of using active control studies to evaluate the effectiveness of investigational new animal drugs. Study design issues will be discussed, including the selection of the control group (approved drugs, standard care, negative, unapproved drugs), the appropriate analysis methods, and verifying the assumptions of a valid non-inferiority comparison. The implications of existing draft guidance such as "Non-Inferiority Clinical Trials", "Antibacterial Drug Products: Use of Non-inferiority Studies to Support Approval", and "Active Controls in Studies to Demonstrate the Effectiveness of a New Drug for Use in Companion Animals" (soon to be published) will be discussed from a clinical and statistical perspective.

Non-inferiority Veterinary Clinical Trials: Design and Analysis Issues
View Presentation View Presentation Virginia F Recta, Food and Drug Administration, CVM

Use of Non-inferiority Studies in Veterinary Medicine from a Clinical Perspective
View Presentation View Presentation Lisa M Troutman, FDA/CVM

Non-Inferiority in Veterinary Medicine - Lessons from the EU
View Presentation View Presentation Guenther Strehlau, Novartis Animal Health

Discussant(s): Terry Settje, Bayer Healthcare


PS3a Adaptive Design Guidance Discussion and Case Sharing

4:15 PM - 5:30 PM
Constitution Room A

Organizer(s): Yeh-Fong Chen, FDA; Weili He, Merck & Co., Inc.; Yannis Jemiai, Cytel, Inc.

Chair(s): Weili He, Merck & Co., Inc.

In response to the FDA Critical Path Initiative and to increase efficiency of the drug development process with the aim of bringing new drugs to the market quicker, pharmaceutical companies have considered implementing adaptive designs in clinical trials. The statistical properties and theoretical gains provided by adaptive designs have been studied extensively. Nevertheless, to completely understand the essence and practical use of adaptive designs, it is critical to get familiarized with the regulatory guidance and learn about their applicability to real case examples. In this session, one speaker from FDA will share FDA's perspective as laid out in the adaptive design guidance. Two speakers from industry will then describe real case studies in applying adaptive designs. Tying it all together, another member from industry will discuss the industry’s perspective and reactions to the adaptive design guidance using the case studies as examples, to outline the issues and obstacles the industry still faces in conducting adaptive design clinical trials.

Regulatory Perspectives of Adaptive Design Throughout Drug Developments
Sue-Jane Wang, FDA

An Adaptive Design for Case-Driven Efficacy Study When Incidence Rate is Unknown – A Case Study
Xiaoming Li, Merck Research Laboratories

A Sample of Adaptive Dose-Finding Case Studies
View Presentation View Presentation Brenda L Gaydos, Eli Lilly and Company

Discussant(s): Keaven Anderson, Merck & Co., Inc.


PS3b Meta-Analysis in Regulatory Practice

4:15 PM - 5:30 PM
Constitution Room B

Organizer(s): Xiao Ding, Merck & Co., Inc.; Susan Huyck, Merck & Co., Inc.; Laura Lu, FDA; Deborah Shapiro, Merck & Co., Inc.

Chair(s): Laura Lu, FDA

In regulatory practice, meta-analysis is an approach for drawing conclusions regarding drug efficacy and/or safety based on information from multiple clinical trials. Areas in which meta-analysis have been used include, but are not limited to, safety and efficacy labeling updates for a drug or drug class, for a subgroup of patients, biomarker identification, surrogate endpoint qualification, and for characterizing the previous effect of the active control in non-inferiority trials. In this session, we will present examples of meta-analyses in some of these areas, illustrating the challenges and cautions when planning and conducting a meta-analysis in a regulatory setting. Examples will include establishing non-inferiority in HIV trials where the active control effects over placebo may not be directly available because the active control itself may be established from non-inferiority trials, an updated meta-analysis of rosiglitazone trials evaluating cardiovascular safety, and a demonstration of some of the challenges of meta-analysis using examples with SSRIs, anti-psychotics in elderly patients, and anti-epileptic drug analyses.

Meta-analysis in regulatory practice
Tom Hammerstrom, FDA

Meta-Analysis of Rosiglitazone
Fiona M Callaghan, FDA White Oak

Use of meta-analysis in support of product label changes
View Presentation View Presentation Jesse Aaron Berlin, Johnson & Johnson Pharmaceutical Research and Development

Discussant(s): LaRee Tracy, FDA


PS3c Analytical Methods in Drug Safety Signal Detection Research

4:15 PM - 5:30 PM
Constitution Room C

Organizer(s): James J. Chen, FDA/NCTR; Wei Deng, Novartis Pharmaceuticals; Stephanie Keeton, FDA; Wei Wang, Eli Lilly and Company

Chair(s): Wei Wang, Eli Lilly and Company

The increased focus on the safety of medical products, as well as the growing volume of available safety information, has created a need for objective quantitative approaches to supplement the medical review of individual case safety reports. Various analytical methods have been used to identify trends and relationships in both clinical and post-marketing safety databases in support of safety signal detection. In this session, we will introduce some cutting edge analytical methods for safety signal detection as well as some powerful data visualization methods that are extremely valuable to facilitate the medical review of complex information and maximize the ability to detect any unexpected and unusual features.

Statistically-Guided Review of Safety Data in Clinical Trials
Michael O'Connell, Spotfire, TIBCO Software, Inc

Graphical Approaches for Safety Signal Detection - Efforts of the FDA-PhRMA-Academia Working Group
View Presentation View Presentation Mat Soukup, U.S. Food and Drug Administration

Data Mining Methods for Clustering Large Two-way Data to Identify Local Structures and Global Patterns
Minho Chae, FDA


PS3d Challenging Statistical Issues in Design and Analysis of Thorough QT/QTc Studies

4:15 PM - 5:30 PM
Constitution Room D

Organizer(s): Hui Quan, Sanofi-Aventis

Chair(s): Joanne Zhang, FDA

ICH region regulatory agencies request all sponsors of new non-arrhythmic drugs to conduct thorough QT/QTc studies to determine whether the drugs have potential effects to delay cardiac repolarization measured by QT/QTc interval. Although the ICH E14 (2005) guidance provides useful recommendations to the industry, there are still multiple remaining challenging statistical issues related to efficient QTc trial design and data analysis. They include (1) sample size considerations for different types of thorough QT/QTc study designs; (2) methods for baseline covariate adjustment and the corresponding impact on variability; (3) conservativeness of the conventional primary intersection-union test for assessing active treatment effect; (4) multiplicity adjustments for demonstrating assay sensitivity or known effect of the active control; and (5) PK/PD modeling for connecting drug-exposure and QTc response. In this session, presenters from the FDA, academia, and industry will have the opportunity to share their thoughts on addressing these issues based on their research and trial experience in the area.

Comparison of statistical models adjusting for baseline in the analysis of parallel-group thorough QT/QTc studies
Guowen (Gordon) Sun, sanofi-aventis US

Exposure-Response Modeling Approach for Assessing QT Effect in 'Thorough' QT/QTc Studies
View Presentation View Presentation Balakrishna Sadashiv Hosmane, Northern Illinois University

Mean or Time-matched? The Choices of Baseline QTc for Parallel TQT Studies
View Presentation View Presentation Qianyu Dang, FDA

Discussant(s): Yi Tsong, FDA


PS3e Analysis of Safety Data Collected from Veterinary Efficacy Studies

4:15 PM - 5:30 PM
Constitution Room E

Organizer(s): Todd Blessinger, FDA/CVM; Junshan Qiu, FDA/CVM; Steven Radecki, Industry Consultant

Chair(s): Todd Blessinger, FDA/CVM

In evaluating veterinary drugs for approval, safety data are collected from both target animal safety studies and field effectiveness studies. Target animal safety (TAS) studies provide data from exaggerated dosing in healthy subjects, while field effectiveness studies provide safety data under actual field conditions of use. The session will focus on the safety data collected from the field effectiveness studies. These data can provide substantial safety information for veterinary drug reviewers given the larger number of animals enrolled (compared to TAS studies) and the administration of the drug under field conditions. Field effectiveness studies are diverse and employ many different designs, so safety data collected from these studies may need to be analyzed in many different ways. The session will have speakers from both FDA and industry sharing their expertise and facilitating discussion on the type and quality of safety information generated from field studies and appropriate ways to analyze, present, and interpret these diverse data sets.

Analysis of safety data collected from veterinary effectiveness studies.
View Presentation View Presentation Michele Sharkey, FDA/CVM

Safety Data from Field Efficacy Studies: An Industry Perspective
View Presentation View Presentation Vickie L. King, Pfizer Animal Health

Graphical Methods for Safety Data from Efficacy Clinical Trials
View Presentation View Presentation Vicki Ann Lancaster, FDA/CVM/ONADE/DSS


Wed, Sep 22

CMC 1 Basis and Framework for Setting Meaningful Decision Criteria to Ensure Product Quality

8:30 AM - 9:45 AM
Constitution Room E

Organizer(s): Anthony Lonardo, ImClone Systems; Jinglin Zhong, FDA/CDER

Chair(s): Anthony Lonardo, ImClone Systems

The FDA publication of quality by design (QbD) initiatives in 2006 represented a significant change and evolution in how pharmaceutical companies achieved and demonstrated product quality. Central to the implementation of QbD is the determination of measurable operational limits for critical process and product attributes, analytical method properties, and post licensure product and analytical characteristics, within which it is believed the product will be fit for use. Statisticians and subject matter experts work to craft objective decision criteria (acceptance criteria) to be applied at the most critical times during a product’s lifecycle. The criteria could be based on clinical considerations, scientific principles, biological models, e.g. for MOA or PK, experience with similar products, or process capability. This session will explore considerations for a framework for setting meaningful acceptance criteria, including issues related to limited data and possible use of historical and development experience. Some of the questions which will be addressed include: How do we establish the link between acceptance limits and product quality? How do we operationally quantify and balance risk and practicality in setting or modifying acceptance limits? How should we manage the desire to establish acceptance criteria with limited data? If not, can one use other data, e.g. from similar processes/products, to augment the data to obtain meaningful limits? How? Should there be a mechanism to update and improve the estimates as more information becomes available? Are there flexible approaches to reduce the overhead of regulatory burden?

The Bases for Establishing Acceptance Criteria in CMC Applications
View Presentation View Presentation Tim Schofield, GlasxoSmithKline

Setting Acceptance Limits – Equivalence Limits for Assessing Similarity as an Example
View Presentation View Presentation Walter W. Hauck, US Pharmacopeia

FDA Team Panel
Yi Tsong, FDA


PS4a Issues and Suggestions Involving Subgroup Analyses

8:30 AM - 9:45 AM
Constitution Room A

Organizer(s): Susan Duke, GlaxoSmithKline; David Hebert, UCB Inc.; Mark Rothmann, FDA; Jenny Zhang, FDA

Chair(s): Mark Rothmann, FDA

While the primary analysis is generally based on the intention-to-treat population, many subgroup analyses are performed and their results may impact regulatory decisions. Subgroup analyses may be done to investigate the robustness, the internal consistency of effects, to confirm the overall conclusions of a trial, in a risk-benefit assessment of which populations may be most likely benefit from the drug, to generate hypotheses to be tested in future clinical trials, to suggest adaptations to an ongoing trial and as a primary analysis in a clinical trial. When the overall results do not reach statistical significance, the results found in subgroup analyses can be controversial. It should be recognized that subgroup or interaction analyses are generally exploratory and that chance can lead to large differences in the estimated effects across investigated subgroups. In this session, the speakers will focus on the issues and controversies involving subgroup analyses.

Subgroup Analysis in Clinical Trials
View Presentation View Presentation Rui Wang, Massachusetts General Hospital and Harvard Medical School

Subgroups analyses – why and how
View Presentation View Presentation Brian Wiens, Alcon

A general approach for testing a prespecifeid subgroup in clinical trials
Mohamed Ahmad Alosh, FDA

Discussant(s): Thomas Fleming, University of Washington


PS4b Critical Statistical Issues and Standards for Biomarker Discovery and Diagnostic Assay Development

8:30 AM - 9:45 AM
Constitution Room B

Organizer(s): Hope Knuckles, Abbott; Jae Lee, University of Virginia, School of Medicine; Kyunghee Song, FDA

Chair(s): Jae Lee, University of Virginia, School of Medicine

Over the past few years, there have been many published claims for diagnostic biomarkers derived from high throughput biological assays. Companies have been formed, and tests advertised. These claims, however, have not led to many changes in patient care, due in large part to poor validation performance. One reason for this disconnect is that there is currently little consensus as to the minimal statistical standards that must be followed in biomarker assay development. Some improvements have been made, such as the MIAME standards for microarray data, the REMARK guidelines for reporting standards (McShane, et. al.) and the FDA's draft guidance for in-vitro diagnostic multivariate index assays (IVDMIAs), but these are empirically not enough to allow for analytical reproducibility and have not been universally adopted by journals. In this session, we expand on some of the issues already identified and attempt to extend these earlier efforts to provide a broader set of Minimal Statistical Standards for Biomarker Assay Development. Since this session will be a concurrent session, each speaker will be given 20 minutes to present, and the discussant will be given 15 minutes to lead a discussion. A round table is planned to compliment this session and to provide more needed discussion on this topic

Statistical Challenges in the Development and Validation of Diagnostic Tests: A Regulatory Perspective
View Presentation View Presentation Gregory Campbell, U.S. Food and Drug Administration

Industry Demands and Statistical Perspective for Drug Biomarker Development
View Presentation View Presentation Kwan Lee, GSK

Reproducible Research: Case Studies in Forensic Bioinformatics
View Presentation View Presentation Keith Baggerly, UT M.D. Anderson Cancer Center

Discussant(s): Lisa McShane, National Cancer Institute


PS4c Recent Issues in Diabetes Drug Development

8:30 AM - 9:45 AM
Constitution Room C

Organizer(s): Alan Chiang, Eli Lilly; Janice Derr, FDA; Feng Gao, GlaxoSmithKline

Chair(s): Alan Chiang, Eli Lilly

The final guidance document "Diabetes Mellitus: Evaluating Cardiovascular Risk in New Antidiabetic Therapies to Treat Type 2 Diabetes" issued by the FDA in December 2008 provided recommendations to sponsors on how to evaluate cardiovascular risk for a new antidiabetic therapy for treatment of type 2 diabetes. Specifically, this document describes the need for meta-analyses of cardiovascular endpoints in phase 2 and phase 3 clinical trials and for the possible need of a large cardiovascular outcomes study. Approaches to the design of the phase 2 and 3 studies to allow for a meta-analysis of this data and methods for conducting the meta-analysis will be presented. In addition, design and analysis of the large cardiovascular outcomes study to meet the guidance thresholds around cardiovascular risk will be discussed. This session will provide an opportunity to discuss these important issues and share key learning.

Bayesian Adaptive Designs for FDA Required Safety Studies
View Presentation View Presentation Scott M Berry, Berry Consultants

Assessing Cardiovascular Risk in Diabetes Drug Development
View Presentation View Presentation David H Manner, Eli Lilly and Company

Statistical Implementation of the Diabetes Guidance for Cardiovascular Outcomes
David Hoberman, Food and Drug Administration

Discussant(s): Jesse Aaron Berlin, Johnson & Johnson Pharmaceutical Research and Development


PS4d Statistical Methods for Efficacy Analysis in the Presence of Substantial Patient Dropout in CNS Trials

8:30 AM - 9:45 AM
Constitution Room D

Organizer(s): Suna Barlas, Pfizer; Joan Kempthorne-Rawson, Boehringer Ingelheim; Fanhui Kong, FDA

Chair(s): Yeh-Fong Chen, FDA

Patient dropout in the clinical trials of the CNS therapeutic area is particularly rampant. The dropout rate could reach 60% or higher in some schizophrenia trials. With high dropout percentages in categories such as lack-of-efficacy, adverse effect, etc., missing mechanisms could be complicated and could also be highly correlated with the outcome variables causing non-ignorable dropout. Current statistical methodologies for efficacy analysis based on assumptions such as LOCF, MAR or some kind of MNAR models have not fully addressed the complexity of patient dropout in this area. In this session, the repeated measurements and time to discontinuation will be jointly analyzed using shared parameter random effects models to tackle non-ignorable dropout; different patterns of patient response profiles in schizophrenia trials will be graphically compared and the use of time to discontinuation as an outcome variable will be discussed. Finally, the integrated FDA schizophrenia trials database will be used to evaluate reliable statistical methods for efficacy analysis.

Non-ignorable Dropout in Clinical Trials – Case Studies
View Presentation View Presentation Edward F Vonesh, Northwestern University

Evaluation of Treatment Discontinuation in the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) Schizophrenia Study
View Presentation View Presentation Sonia M Davis, Quintiles, Inc.

Exploring the dropout patterns and characteristics in schizophrenia trials
View Presentation View Presentation Phillip Dinh, Food and Drug Administration


CMC 2 Aspects of QbD for Quality Assays and Meaningful Decision Limits

10:00 AM - 11:15 AM
Constitution Room E

Organizer(s): Anthony Lonardo, ImClone Systems; Jinglin Zhong, FDA/CDER

Chair(s): Harry Yang, MedImmune

As pharmaceutical R&D is becoming globalized, there is a growing need to develop, validate and transfer assays to multiple sites for clinical testing and product release. However, setting acceptance limits that reflect and balance industry and regulatory risks is a continuing challenge. It is also unclear if the tolerance to risk and its control is or should be the same during all stages of assay development. This session will explore the key challenges associated with developing quality assays and setting meaningful decision limits.

A novel bootstrap approach for setting acceptance criteria on potency assay parameters
View Presentation View Presentation Daniel Joelsson, Merck & Co. Inc; Yue Wang, Merck & Co. Inc

QbD for Biotechnology Products, Where are We Now?
Kathy Lee, FDA

CMC2 Panel
Stan Altan, Johnson & Johnson; Walter W. Hauck, US Pharmacopeia; Kathy Lee, FDA; Tsai Lien, FDA/CBER; Jinglin Zhong, FDA/CDER; Tim Schofield, GlasxoSmithKline


PS5a Survival Analysis: Statistical Challenges and Special Methods

10:00 AM - 11:15 AM
Constitution Room A

Organizer(s): Yoko Adachi, U.S. Food and Drug Administration Center for Veterinary Medicine; Qing Xu, FDA; Xiongce Zhao, NIH; John Zhong, Human Genome

Chair(s): Qing Xu, FDA

Survival analysis presents many important and unique challenges to the design and the evaluation of efficacy and safety across all phases of clinical development. Special methodologies for survival analysis are required to solve particular problems that arrive from different type of clinical trials. In this session, experts from academia, industry and regulatory agencies will present various issues and methods in survival analysis. Regulatory prospective and suggestions for the design of future studies will be discussed. Special topics include methods for the analysis of interval-censored survival data, methods to address sample size estimation for recurrent events in survival analysis, and use of quintiles in survival analysis. Examples from typical clinical trials will be presented to highlight these issues.

Methods for the Analysis of Interval-Censored Survival Data
View Presentation View Presentation Elizabeth C. Wright, NIDDK/NIH

An alternative summary measure for time-to-event data: residual life?
View Presentation View Presentation Jong-Hyeon Jeong, University of Pittsburgh

Sample Size Estimation for Trials With Recurrent Events as the Primary Endpoint
View Presentation View Presentation Kuolung Hu, Amgen

Discussant(s): Raji Sridhara, FDA


PS5b Missing Data in Clinical Trials with Examples from Oncology and Infectious Diseases

10:00 AM - 11:15 AM
Constitution Room B

Organizer(s): Kyung Yul Lee, FDA; Margaret Minkwitz, AstraZeneca; Yuan-Li Shen, FDA; Shailaja Suryawanshi, Merck & Co., Inc.

Chair(s): Kyung Yul Lee, FDA

In clinical trials, missing data due to dropouts or lost-to follow-ups leads to bias and variability in estimation and inference and hence impacts regulatory decision making. To meaningfully reduce the level of missing data, it is important to recognize and address many factors that commonly lead to higher levels of missingness. This section will discuss some perspectives on addressing missing data issues through examples in areas of oncology and infectious diseases. In oncology trials, it is very difficult to have a meaningful intent-to-treat analysis for a regulatory endpoint of progression-free survival due to extensive loss to follow-up. The censoring due to lost to follow-up can be informative and introduce bias. Statistical analyses when there is extensive informative censoring are not valid and make it difficult to interpret results. Pros and Cons of several proposed sensitivity analyses and imputation methods used to support the robustness of conclusions drawn from the primary analysis will be discussed.

Addressing Missing Data In Clinical Trials
Thomas Fleming, University of Washington

Examining the informativeness of premature censoring in evaluating progression-free survival
View Presentation View Presentation Mark Rothmann, FDA

Analyzing Progression Free Survival: A Meta-analysis of 28 Trials to Explore the Impact of Censoring Rules.
View Presentation View Presentation Jonathan Denne, Eli Lilly


PS5c Heterogeneity in Treatment Response: What It Is, Why It's Important and How to Find It

10:00 AM - 11:15 AM
Constitution Room C

Organizer(s): Joe Cappelleri, Pfizer; Yuying Jin, FDA; Yu (Audrey) Zhao, FDA/CDRH

Chair(s): Joe Cappelleri, Pfizer

Heterogeneity in treatment response is encountered in clinical trials due to site variation or factors outside the context of pre-defined subgroups. Analytic methods are now available to identify individuals who are more or less responsive to treatment in clinical trials using appropriate responder thresholds. Once identified, these individuals can then be assessed to see if they have common characteristics that may shed light on their differential response to treatment. For identifying sources of heterogeneity, a method of model expansion and factor mixture model (FMM) will be considered. The former addresses modeling heterogeneity across the study sites by treating the site effects as random and using the log-linear variance function. FMM is designed for data that appears heterogeneous but may include several latent classes that are homogeneous. It combines latent class analysis and common factor analysis, and is a good choice if it is reasonable to assume that observed variables within each class can be modeled using a common factor model.

Heterogeneity issue in clinical trial
View Presentation View Presentation Chul Ahn, FDA/CDRH

Examining heterogeneity in treatment response in clinical trials: The poor man’s mxture model analysis
View Presentation View Presentation Kathleen W. Wyrwich, UnitedBioSource Corporation

Examining heterogeneity in treatment response in clinical trials: Use of factor mixture models
View Presentation View Presentation Donald E Stull, United BioSource Corporation

Discussant(s): Ari Gnanasakthy, Novartis


PS5d Emerging Statistical Methods in Preclinical Safety Assessment

10:00 AM - 11:15 AM
Constitution Room D

Organizer(s): James J. Chen, FDA/NCTR; C. Thomas Lin, Abbott Lab; Karl Lin, FDA/CDER; Maria Mendoza, FDA/NCTR

Chair(s): James J. Chen, FDA/NCTR

Preclinical studies are conducted for safety assessment from exposure to a compound or drug. Types of outcomes range anywhere from adverse effects such as the interference with natural physiological responses to the development of toxicological endpoints, such as drug-induced cardiac liabilities, tumor formation or non-genotoxic reproductive effects. Development of prediction model for biomarker identification and statistical methods for analyzing preclinical data to address these types of emerging concerns is the focus of this session.

Developing Genomic Biomarkers for Early Safety Screens for Non-genotoxic Carcinogenicity
Nandini Raghavan, Johnson & Johnson Pharmaceutical Research & Development

A Quasi-Likelihood Model for the Analysis of Tumorigenicity in Multigenerational Data
Maria Mendoza, FDA/NCTR

Prediction of In Vivo Toxicity Endpoints Using In Vitro Bioassay and Numerical Descriptors
Stanley Young, National Institute of Statistical Sciences


CMC 3 Comparability and Stability Decision Limits within the QbD Paradigm

12:45 PM - 2:00 PM
Constitution Room E

Organizer(s): Anthony Lonardo, ImClone Systems; Jinglin Zhong, FDA/CDER

Chair(s): Jinglin Zhong, FDA/CDER

In this session we will deal with two important decision points that could have significant impact on product quality. Comparability/Design Space Pro/Con from the perspective of the reviewer: When there is a change to a process, it is required by regulations that a study be conducted to establish comparability between product from the new and old processes. The FDA has provided guidance on comparability which lays out the objectives for a protocol. Under QbD, the definition of design space provides one mechanism to permit process change without a formal comparability study or regulatory supplement. However there are times when changes may be required which were not considered during the original development of the design space. Oftentimes, the evaluations are conducted through statistical evaluation. Key to a statistical approach is the formulation of an appropriate research hypothesis. While many statisticians agree that this should be formulated as an equivalence hypothesis, there is little guidance regarding an appropriate equivalence criterion? This session will address the unique challenges associated with establishing decision criteria to determine an acceptable or comparable process or product. Shelf Life: ICH Q1E(2003)recommends a confidence interval approach for determining shelf life of a pharmaceutical product. Various other methods have also been proposed to address what some perceive as shortcomings of the ICH recommendation. What is lacking is a common framework to evaluate, compare and quantify risks and benefits. A unified risk based method in the spirit of QbD will be presented and discussed in which design considerations arising from the Design Space will be related to shelf life impact considerations.

Design Space Development and Verification
View Presentation View Presentation Christine M.V. Moore, FDA/CDER/ONDQA

A Risk-based Approach for Determination of Shelf Life of Pharmaceutical Products
Harry Yang, MedImmune

CMC3 Panel
James Schwenke, Boehringer Ingelheim; Arzu Selen, FDA; Tim Schofield, GlasxoSmithKline; Meiyu Shen, FDA/CDER; David Christopher, Merck & Co., Inc.


PS6a Recent Developments in Non-Inferiority Trials – Regulatory, Design and Analysis

12:45 PM - 2:00 PM
Constitution Room A

Organizer(s): Somesh Chattopadhyay, FDA; George Chi, J&JPRD; Kalyan Ghosh, BMS; Ram Tiwari, FDA

Chair(s): George Chi, J&JPRD

EMEA has published in 2005, the Points-to-Consider document on non-inferiority trials, while FDA has just released its draft Guidance on non-inferiority trials in March, 2010. This session has invited two key speakers from each regulatory body to discuss the current regulatory guidelines and views on the design, analysis and interpretation of non-inferiority trials. They will also discuss some of the issues that are still confronting the regulatory agencies on non-inferiority trials and provide their recommendations on how to deal with these issues. A new perspective on the design and analysis of non-inferiority trials will be discussed and presented by a third speaker. This perspective is based on recent research results and will have important and far-reaching implications on the design, analysis and interpretation of non-inferiority trials. A new integrated and more satisfactory approach towards non-inferiority trials will become apparent

Statistical Considerations in Design and Analysis of Active Controlled Clinical Trials
View Presentation View Presentation Hsien (James) Hung, FDA

Non-Inferiority Trials: Issues from an Academic’s View
Ralph D'Agostino, Sr., Boston University

Inferiority Index and Non-inferiority Trials
Gang Li, J&J PRD


PS6b Statistical Issues Related to Development and Use of Predictive Biomarkers

12:45 PM - 2:00 PM
Constitution Room B

Organizer(s): Rebecca Hozak, Eli Lilly; Deepak Khatry, MedImmune; Estelle Russek-Cohen, FDA; Yang Yang, FDA

Chair(s): Rebecca Hozak, Eli Lilly

Biomarkers are increasingly being pursued as a means to select optimal treatment for patients, in regard to both efficacy and safety. It is hoped that the use of such predictive biomarkers will also increase efficiency of research and development of new therapies. In the last few years, studies have demonstrated progress in identification and clinical validation/qualification of predictive biomarkers. This session will bring together a panel of prominent statisticians from the FDA, industry, and non-profit research to share their current state of knowledge on the promises and limitations of using predictive biomarkers. The speakers will focus on study designs, statistical requirements, and regulatory expectations on the path to developing predictive biomarkers. Following the presentations, questions will be posed to the speakers and an open floor discussion will be held.

Statistical Issues Related to Development and Use of Predictive Biomarkers
Sue-Jane Wang, FDA

Design Principles for predictive marker validation
View Presentation View Presentation Daniel J Sargent, Mayo Clinic

Issues for the Industry Statistician When a Potential Predictive Biomarker is Identified
View Presentation View Presentation Nancy Gustafson, Bristol-Myers Squibb Co.


PS6c Quantitative Assessments of Patient Reported Outcomes: Questionnaire Design and Analysis

12:45 PM - 2:00 PM
Constitution Room C

Organizer(s): Rima Izem, FDA/CDER/OB/DB4; Tammy Massie, FDA/CBER; Margaret Rothman, Johnson & Johnson

Chair(s): Rima Izem, FDA/CDER/OB/DB4

The session and panel speakers will discuss the role of statisticians in both the development and the use of Patient Reported Outcome instruments (PRO) in clinical studies. The statistician’ role in developing or evaluating a PRO used as a key endpoint for labeling is important in light of the recently issued PRO guidance for industry (December 2009). This guidance sets a framework for facilitating discussion of issues related to PRO's. The guidance recommends the use of psychometric tools (such as Factor Analysis and Rasch analysis) to quantitatively assess the validity and interpret PRO scores. However, the regulatory implications of these tools’ results in clinical trials are not well established. Speakers will focus on design and quantitative methods in the PRO instrument development and review process such as: study design, item selection, questionnaire validation and validation of clinical difference in PRO scores. They will discuss regulatory challenges these methods present and propose improvement to the process.

What Industry Statisticians Should Know about Patient-Reported Outcomes to Support Label Claims
View Presentation View Presentation Joe Cappelleri, Pfizer

Challenges facing FDA statisticians in their reviews of submissions containing patient-reported outcomes (PROs)
View Presentation View Presentation Lisa A. Kammerman, FDA

Discussant(s): Rima Izem, FDA/CDER/OB/DB4; Margaret Rothman, Johnson & Johnson; Kathleen W. Wyrwich, UnitedBioSource Corporation


PS6d Statistical Challenges Unique to Medical Devices

12:45 PM - 2:00 PM
Constitution Room D

Organizer(s): Chul Ahn, FDA/CDRH; Weihua Cao, FDA/CDRH; Peter Lam, Boston Scientific

Chair(s): Chul Ahn, FDA/CDRH

This session will discuss two statistical problems unique to medical device clinical trials. It includes blinding sham control from neuromodulation studies for pain management or migraine, and randomization within subject where effectiveness is measured at vessel level in interventional cardiology studies, but safety is measured at subject level. For each problem, the industry and the FDA perspective of the challenges and opportunities will be discussed.

Placebo Control and Patient Blinding in Clinical Trials of Neurostimulation Devices
Nitzan Mekel-Bobrov, Boston Scientific Neuromodulation

A statistical reviewer’s perspective on using sham as the control in neuromodulation device studies
View Presentation View Presentation Martin Ho, Center for Devices and Radiological Health

Design and Analysis Issues for Interventional Cardiology Studies in Patients with More than One Lesion Treated
View Presentation View Presentation Peggy J Pereda, Boston Scientific Corporation

Discussant(s): Yunling Xu, FDA


PS7a Population Enrichment Designs in Clinical Trials

2:15 PM - 3:30 PM
Constitution Room A

Organizer(s): Brent Burger, Cytel, Inc.; Mohammad Huque, FDA; Chia-Wen Ko, FDA

Chair(s): Fanhui Kong, FDA

Some treatments benefit only a subgroup of patients who receive them. The current study design and analysis paradigm with focus on efficacy of main effects of treatment and all other analyses regarded as exploratory is very inefficient and time consuming in identifying the treatment efficacy of the therapy in the particular subgroup. Population enrichment designs provide an opportunity to adaptively analyze the treatment benefit of specific subgroups of the patient population. In such a design one would prospectively specify the subgroups that the new therapy is expected to benefit. Starting with the initial enrollment from a broad patient population one can select only those subgroups that appear to be benefiting from the experimental therapy based on the results of interim analyses. Such a clinical trial strategy could greatly improve the chances of success for the trial. In this session, new statistical procedures of such designs and their applications in clinical trials will be presented. Statistical, regulatory and logistical issues will be discussed.

Predictive Analysis of Phase III Clinical Trials
View Presentation View Presentation Richard M. Simon, National Cancer Institute

Population Enrichment Designs: Case Study of a Large Multi-national Cardiovascular Trial
View Presentation View Presentation Cyrus R Mehta, Cytel Inc.

Adaptive Patient Population Selection Design in Clinical Trials
View Presentation View Presentation Hui Quan, Sanofi-Aventis

Discussant(s): Sue-Jane Wang, FDA


PS7b Drug Safety in Clinical Trials

2:15 PM - 3:30 PM
Constitution Room B

Organizer(s): Aloka Chakravarty, FDA; Jie Chen, Merck Serono

Chair(s): Aloka Chakravarty, FDA

In the setting of assessing a specific rare and serious adverse event in clinical trials, sample size is often of concern. One approach is to take the continuous sequential monitoring scheme on safety endpoint into account when designing the study in which superiority and non-inferiority hypotheses are formulated on efficacy and safety endpoints and the two-way multiplicity adjustment and the different testing schemes are considered. On the other hand, in cumulative meta-analysis setting the regular statistics may not maintain the correct type 1 error rate. The cumulative meta-analysis using the law of iterated logarithm (LIL) proposed by Lan et al may be considered for safety assessment. Furthermore, protocols that limit enrollment eligibility may introduce selection error that severely limits a RCT's applicability to a wide range of patients. Hence, an estimator of effect size that capitalizes on RCTs' strong internal validity and observational studies' strong external validity is proposed and assessed based on single sources of evidence.

Sequential Generalized Likelihood Ratio Tests for Safety Evaluation
View Presentation View Presentation Jie Chen, Merck Serono

Comparison of the performance of cumulative meta-analytical methods with rare events
View Presentation View Presentation Xiao Ding, Merck & Co., Inc.

Quantifying and Correcting Generalization Bias in Safety Assessment
View Presentation View Presentation Eloise Kaizar, Ohio State University


PS7c Design and Analysis of Clinical Trials with Multiple Endpoints

2:15 PM - 3:30 PM
Constitution Room C

Organizer(s): Ghideon Ghebregiorgis, FDA; Peng Huang, Johns Hopkins University; Scott Miller, FDA; Jodi Rylance, Quintiles, Inc.

Chair(s): Peng Huang, Johns Hopkins University

Treatment comparison in clinical trials often involves the evaluation of multiple endpoints due to the fact that multiple measurements may be needed to characterize a particular condition or disease and the benefit of a treatment. Although a number of methods have been proposed in literature from defining a composite endpoint to analyzing multiple primary endpoints, each of them is best suited to address a certain question. In this session we will discuss various statistical strategies in order to answer the right medical questions; we will illustrate how trial results should be interpreted scientifically; and we will present some of the FDA’s concerns along with some recommendations.

Adaptive Designs for Clinical Trials with Multiple Endpoints
View Presentation View Presentation Ming T Tan, Division of Biostatistics

Improving the Information Content of RCT Endpoints
View Presentation View Presentation Vance Berger, NIH

Changing Paradigms in Cancer Clinical Trials
View Presentation View Presentation Jeanne Kowalski, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins

Discussant(s): Barbara Tilley, University of Texas Health Science Center at Houston


PS7d Causal Inference in Clinical Trials

2:15 PM - 3:30 PM
Constitution Room D

Organizer(s): C.V. Damaraju, Johnson & Johnson; Thamban Valappil, CDER/FDA

Chair(s): C.V. Damaraju, Johnson & Johnson

Randomized clinical trials are ideal designs to evaluate efficacy and safety of outcomes while ensuring comparability at baseline. However, it does not guarantee an unbiased comparison when post-treatment variables have the potential to seriously confound the treatment effects. Confounding due to non-compliance and informative drop-outs can severely limit the ability to draw inferences and sometimes lead to potentially incorrect conclusions. The causal effect of a new treatment compared to an active control is the expected difference between the potential outcomes given new treatment and control. The counterfactual framework and causal modeling using Marginal Structural Models (MSM) can improve efficiency compared to standard approaches for estimating treatment differences. There are a variety of estimators which include inverse weighting, double robust, G-computation and targeted MLE. Sensitivity of these estimators compared to standard approaches may be of great interest. The goal of the session is to motivate and discuss the role of causal inference in clinical trials with the underlying theory and applications

Targeted Maximum Likelihood Based Super Learning: Assessing Effects in RCT and Observational Studies
View Presentation View Presentation Mark van der Laan, University of California, Berkeley

Estimating the Causal Effect of Low Tidal Volume Ventilation on Survival in Patients with Acute Lung Injury
Chenguang Wang, FDA